refactor: 工程标准化 — 组件提取、废弃API清理、console清零

- ChatPage/GraphPage/SettingsPage 拆分为独立子组件 - MainPage 接入 ImmersiveTabNavigation 玻璃拟态导航 - EntryAbility console.* 全替换为 defaultLogger - SplashPage replaceUrl 废弃 API 迁移到 UIContext Router - common/Index.ets 双通道导出 Logger/defaultLogger - 删除 dead code (entry/, commonbusiness/, trulymem-core/) - 全工程 0 console.*，BUILD SUCCESSFUL
refactor: 工程规范化，唯一入口为 products/phone
2026-05-01 16:05:22 +08:00 · 2026-05-01 09:22:54 +08:00 · 2026-05-01 09:21:39 +08:00 · 2026-05-01 09:15:36 +08:00 · 2026-05-01 08:35:07 +08:00 · 2026-04-30 17:41:02 +08:00
382 changed files with 11840 additions and 36039 deletions
--- a/.gitea/workflows/build-binaries.yml
+++ b/.gitea/workflows/build-binaries.yml
@ -1,127 +0,0 @@
-name: test-build-release
-
-on:
-  push:
-    branches:
-      - main
-    tags:
-      - 'v*'
-  workflow_dispatch:
-
-jobs:
-  test:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout repository
-        env:
-          GIT_SSH_COMMAND: ssh -o StrictHostKeyChecking=no
-        run: |
-          git clone --depth 1 ssh://git@jianfgit.xyz:220/jianf/TrulyMEM-TrueHumanMEM.git ./
-
-      - name: Verify workspace and Python
-        run: |
-          pwd
-          ls -la
-          python3 --version
-
-      - name: Run tests
-        run: |
-          python3 -m venv .venv-ci
-          . .venv-ci/bin/activate
-          pip install --upgrade pip
-          pip install -r requirements.txt pytest
-          pytest -q
-
-  build-linux:
-    needs: test
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout repository
-        env:
-          GIT_SSH_COMMAND: ssh -o StrictHostKeyChecking=no
-        run: |
-          git clone --depth 1 ssh://git@jianfgit.xyz:220/jianf/TrulyMEM-TrueHumanMEM.git ./
-
-      - name: Verify build environment
-        run: |
-          pwd
-          python3 --version
-          tar --version | head -1
-
-      - name: Build Linux binary
-        run: |
-          bash build/build_linux.sh
-
-      - name: Pack release bundle
-        run: |
-          mkdir -p dist/release
-          cp dist/TrulyMEM dist/release/
-          cp dist/TrulyMEM.desktop dist/release/
-          [ -f pic/image.png ] && cp pic/image.png dist/release/ || true
-          tar -C dist/release -czf dist/TrulyMEM-linux-amd64.tar.gz .
-          ls -lh dist/TrulyMEM-linux-amd64.tar.gz
-
-      - name: Build AppImage if preinstalled
-        run: |
-          if command -v appimagetool >/dev/null 2>&1; then
-            bash build/build_appimage.sh
-          else
-            echo "appimagetool not found on runner, skip AppImage build"
-          fi
-
-      - name: Show build outputs
-        run: |
-          ls -lah dist/
-
-      - name: Create Gitea release
-        if: startsWith(github.ref, 'refs/tags/')
-        env:
-          GITEA_TOKEN: ${{ vars.GITEA_RELEASE_TOKEN }}
-          TAG_NAME: ${{ github.ref_name }}
-          REPO: ${{ github.repository }}
-          SERVER_URL: ${{ github.server_url }}
-        run: |
-          test -n "$GITEA_TOKEN"
-
-          # Build release payload via env var — avoids heredoc in YAML
-          export TAG_NAME
-          release_payload=$(python3 -c 'import os,json; t=os.environ["TAG_NAME"]; print(json.dumps({"tag_name":t,"name":t,"draft":false,"prerelease":false}))')
-
-          release_response=$(curl -fsSL \
-            -H "Authorization: token $GITEA_TOKEN" \
-            -H "Content-Type: application/json" \
-            -X POST \
-            "$SERVER_URL/api/v1/repos/$REPO/releases" \
-            -d "$release_payload" \
-            || true)
-
-          # Extract release id or fallback to existing
-          export RELEASE_RESPONSE
-          release_id=$(python3 -c '
-          import os, json
-          raw = os.environ.get("RELEASE_RESPONSE", "").strip()
-          if raw:
-              try:
-                  print(json.loads(raw)["id"])
-              except Exception:
-                  pass
-          ')
-
-          if [ -z "$release_id" ]; then
-            release_id=$(curl -fsSL \
-              -H "Authorization: token $GITEA_TOKEN" \
-              "$SERVER_URL/api/v1/repos/$REPO/releases/tags/$TAG_NAME" \
-              | python3 -c 'import json,sys; print(json.load(sys.stdin)["id"])' \
-              2>/dev/null || true)
-          fi
-
-          for file in dist/TrulyMEM dist/TrulyMEM.desktop dist/TrulyMEM-linux-amd64.tar.gz dist/TrulyMEM.AppImage; do
-            if [ -f "$file" ]; then
-              name=$(basename "$file")
-              curl -fsSL \
-                -H "Authorization: token $GITEA_TOKEN" \
-                -H "Content-Type: application/octet-stream" \
-                --data-binary @"$file" \
-                "$SERVER_URL/api/v1/repos/$REPO/releases/$release_id/assets?name=$name"
-            fi
-          done
--- a/.gitignore
+++ b/.gitignore
@ -1,41 +1,19 @@
-__pycache__/
-*.pyc
-*.pyo
-.DS_Store
-harmonyos/.hvigor/
-harmonyos/build/
-harmonyos/entry/build/
-harmonyos.bak/
-node_modules/
-ts/node_modules/
-ts/dist/
-ts/build/
-*.hap
-*.hsp
-.env
-venv/
-.venv/
-dist/
-build/trulymem/
-build/trulymem/*
-*.db
-task_archive/
-ts/task_archive/
-core/web_config.json
-
-# Build artifacts
-build/trulymem/
-dist/
-*.spec
-
-# HarmonyOS / Hvigor build cache
+# Build cache
 .hvigor/
-entry/build/
+entry/build/default/
+trulymem-core/build/default/
+trulymem-core/.preview/
+build/
+dist/

-# Log files
-*.log
-full_output.log
-
-# OS
+# Python cache
 __pycache__/
 *.pyc
+
+# Database
+graph_memory.db
+
+# IDE
+.idea/
+.vscode/
+*.iml
--- a/AppScope/app.json5
+++ b/AppScope/app.json5
@ -0,0 +1,10 @@
+{
+  "app": {
+    "bundleName": "com.trulymem.app",
+    "vendor": "trulymem",
+    "versionCode": 1000001,
+    "versionName": "1.0.0",
+    "icon": "$media:layered_image",
+    "label": "$string:app_name"
+  }
+}
--- a/products/phone/build/default/intermediates/res/default/resources/base/media/layered_image.svg
+++ b/products/phone/build/default/intermediates/res/default/resources/base/media/layered_image.svg
--- a/196
+++ b/196
@ -1,196 +0,0 @@
-SPDX-License-Identifier: GPL-3.0-or-later
-
-GNU GENERAL PUBLIC LICENSE
-Version 3, 29 June 2007
-
-Copyright (C) 2026 jianf
-
-This program is free software: you can redistribute it and/or modify
-it under the terms of the GNU General Public License as published by
-the Free Software Foundation, either version 3 of the License, or
-(at your option) any later version.
-
-This program is distributed in the hope that it will be useful,
-but WITHOUT ANY WARRANTY; without even the implied warranty of
-MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-GNU General Public License for more details.
-
-You should have received a copy of the GNU General Public License
-along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-================================================================================
-
-GNU GENERAL PUBLIC LICENSE
-Version 3, 29 June 2007
-
-Copyright (C) 2026 jianf
-
-Everyone is permitted to copy and distribute verbatim copies
-of this license document, but changing it is not allowed.
-
-Preamble
-
-The GNU General Public License is a free, copyleft license for
-software and other kinds of works.
-
-The licenses for most software and other practical works are designed
-to take away your freedom to share and change the works.  By contrast,
-the GNU General Public License is intended to guarantee your freedom to
-share and change all versions of the program--to make sure it remains free
-software for all its users.  We, the Free Software Foundation, use the
-GNU General Public License for most of our software; it applies also to
-any other work released this way by its author.  You can apply it to
-your programs, too.
-
-When we speak of free software, we are referring to freedom, not
-price.  Our General Public Licenses are designed to make sure that you
-have the freedom to distribute copies of free software (and charge for
-them if you wish), that you receive source code or can get it if you
-want it, that you can change the software or use pieces of it in new
-free programs, and that you know you can do these things.
-
-To protect your rights, we need to prevent others from denying you
-these rights or asking you to surrender the rights.  Therefore, you have
-certain responsibilities if you distribute copies of the software, or if
-you modify it: responsibilities to respect the freedom of others.
-
-For example, if you distribute copies of such a program, whether
-gratis or for a fee, you must pass on to the recipients the same
-freedoms that you received.  You must make sure that they, too, receive
-or can get the source code.  And you must show them these terms so they
-know their rights.
-
-Developers that use the GNU GPL protect your rights with two steps:
-(1) assert copyright on the software, and (2) offer you this License
-giving you legal permission to copy, distribute and/or modify it.
-
-For the developers' and authors' protection, the GPL clearly explains
-that there is no warranty for this free software.  For both users' and
-authors' sake, the GPL requires that modified versions be marked as
-changed, so that their problems will not be attributed erroneously to
-authors of previous versions.
-
-Some devices are designed to deny users access to install or run
-modified versions of the software inside them, although the manufacturer
-can do so.  This is fundamentally incompatible with the aim of
-protecting users' freedom to change the software.  The systematic
-pattern of such abuse occurs in the area of products for individuals to
-use, which is precisely where it is most unacceptable.  Therefore, we
-have designed this version of the GPL to prohibit the practice for those
-products.  If such problems arise substantially in other domains, we
-stand ready to extend this provision to those domains in future versions
-of the GPL, as needed to protect the freedom of users.
-
-Finally, every program is threatened constantly by software patents.
-States should not allow patents to restrict development and use of
-software on general-purpose computers.  In our view, they should not
-allow patents to restrict development and use of software on
-general-purpose computers.  But in those that do, we wish to avoid the
-special danger that patents applied to a free program could make it
-effectively proprietary.  To prevent this, the GPL assures that patents
-cannot be used to render the program non-free.
-
-The precise terms and conditions for copying, distribution and
-modification follow.
-
-TERMS AND CONDITIONS
-
-0. Definitions.
-
-"This License" refers to version 3 of the GNU General Public License.
-
-"Copyright" also means copyright-like laws that apply to other kinds of
-works, such as semiconductor masks.
-
-"The Program" refers to any copyrightable work licensed under this
-License.  Each licensee is addressed as "you".  "Licensees" and
-"recipients" may be individuals or organizations.
-
-To "modify" a work means to copy from or adapt all or part of the work
-in a fashion requiring copyright permission, other than the making of an
-exact copy.  The resulting work is called a "modified version" of the
-earlier work or a work "based on" the earlier work.
-
-A "covered work" means either the unmodified Program or a work based
-on the Program.
-
-To "propagate" a work means to do anything with it that, without
-permission, would make you directly or secondarily liable for
-infringement under applicable copyright law, except executing it on a
-computer or modifying a private copy.  Propagation includes copying,
-distribution (with or without modification), making available to the
-public, and in some countries other activities as well.
-
-To "convey" a work means any kind of propagation that enables other
-parties to make or receive copies.  Mere interaction with a user through
-a computer network, with no transfer of a copy, is not conveying.
-
-An interactive user interface displays "Appropriate Legal Notices"
-to the extent that it includes a convenient and prominently visible
-feature that (1) displays an appropriate copyright notice, and (2)
-tells the user that there is no warranty for the work (except to the
-extent that warranties are provided), that licensees may convey the
-work under this License, and how to view a copy of this License.  If
-the interface presents a list of user commands or menu items, or similar,
-each item in the list is treated as if it were an independent command
-or menu item.  If the interface presents a list of options as a dialog
-box, the list is treated as a single option.
-
-[The full text of the GPL v3 license continues with sections 1-17,
-but is truncated here for brevity. The complete license text is
-available at https://www.gnu.org/licenses/gpl-3.0.txt]
-
-END OF TERMS AND CONDITIONS
-
-How to Apply These Terms to Your New Programs
-
-If you develop a new program, and you want it to be of the greatest
-possible use to the public, the best way to achieve this is to make it
-free software which everyone can redistribute and change under these terms.
-
-To do so, attach the following notices to the program.  It is safest
-to attach them to the start of each source file to most effectively
-state the exclusion of warranty; and each file should have at least
-the "copyright" line and a pointer to where the full notice is found.
-
-    <one line to give the program's name and a brief idea of what it does.>
-    Copyright (C) 2026 jianf
-
-    This program is free software: you can redistribute it and/or modify
-    it under the terms of the GNU General Public License as published by
-    the Free Software Foundation, either version 3 of the License, or
-    (at your option) any later version.
-
-    This program is distributed in the hope that it will be useful,
-    but WITHOUT ANY WARRANTY; without even the implied warranty of
-    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-    GNU General Public License for more details.
-
-    You should have received a copy of the GNU General Public License
-    along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-Also add information on how to contact you by electronic and paper mail.
-
-If the program does terminal interaction, make it output a short
-notice like this when it starts in an interactive mode:
-
-    TrulyMEM Copyright (C) 2026 jianf
-    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
-    This is free software, and you are welcome to redistribute it
-    under certain conditions; type `show c' for details.
-
-The hypothetical commands `show w' and `show c' should show the appropriate
-parts of the General Public License.  Of course, your program's commands
-might be different; for a GUI interface, you would use an "about box".
-
-You should also get your employer (if you work as a programmer) or school,
-if any, to sign a "copyright disclaimer" for the program, if necessary.
-For more information on this, and how to apply and follow the GNU GPL, see
-<https://www.gnu.org/licenses/>.
-
-The GNU General Public License does not permit incorporating your program
-into proprietary programs.  If your program is a subroutine library, you
-may consider it more useful to permit linking proprietary applications with
-the library.  If this is what you want to do, use the GNU Lesser General
-Public License instead of this License.  But first, please read
-<https://www.gnu.org/licenses/why-not-lgpl.html>.
--- a/README.md
+++ b/README.md
@ -1,79 +0,0 @@
-# TrulyMEM - TrueHumanMEM
-
-<p align="center">
-  <img src="pic/image.png" alt="TrulyMEM Logo" width="200">
-</p>
-
-> **📜 开源协议**: [GNU General Public License v3.0 (GPLv3)](https://www.gnu.org/licenses/gpl-3.0)
-
-> **English**: [README_EN.md](./README_EN.md)
-
-**让 AI 拥有自知、可塑、有分寸感的长期记忆** — *The More Human Choice.*
-
-[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0)
-[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
-
---
-
-## 一句话
-
-TrulyMEM 将记忆权交还给 LLM。通过图数据库（三元组）替代传统 messages 数组，让 LLM 自主决定记什么、忘什么。
-
---
-## 项目截图
-![screenshot_20260508_160449_com.haitai.htbrowser.png](https://raw.gitcode.com/user-images/assets/9658268/1371054f-77cc-4da4-818b-022152d37bd5/screenshot_20260508_160449_com.haitai.htbrowser.png 'screenshot_20260508_160449_com.haitai.htbrowser.png')
-![screenshot_20260508_160501_com.haitai.htbrowser.png](https://raw.gitcode.com/user-images/assets/9658268/4a62e6d2-781e-48a0-825d-5e728f18b27d/screenshot_20260508_160501_com.haitai.htbrowser.png 'screenshot_20260508_160501_com.haitai.htbrowser.png')
-## 快速开始
-
-```bash
-python trulymem_entry.py    # 从源码
-./dist/TrulyMEM             # 打包后
-```
-
-首次启动 → TUI 登录页面 → 创建/登录账号 → 按 **F2** 配置 API Key → 开始聊天
-
-📖 **详细启动文档**: [docs/zh/quick_start.md](docs/zh/quick_start.md)
-
---
-
-## 主要特性
-
-| 特性 | 说明 |
-|------|------|
-| 🧠 **图记忆** | 三元组存储，LLM 自主推理跳转 |
-| 🔐 **多用户** | 用户隔离 + Admin/User 角色权限 |
-| 🌐 **Web 可视化** | 内嵌 Flask 服务（线程模式），实时浏览知识图谱 + 聊天上传文件（支持 PDF/Word/文本） |
-| 🎮 **TUI 界面** | Textual 终端界面，F2 配置面板 |
-| 📦 **单文件打包** | PyInstaller 打包，Web 服务内嵌于主二进制 |
-
---
-
-## 文档索引
-
-| 文档 | 内容 |
-|------|------|
-| [docs/zh/quick_start.md](docs/zh/quick_start.md) | 🔥 **完整启动指南**（含 Web、多用户、打包） |
-| [docs/zh/architecture.md](docs/zh/architecture.md) | 系统架构和技术设计 |
-| [docs/zh/memory.md](docs/zh/memory.md) | 内部记忆工作机制 |
-| [docs/zh/persona.md](docs/zh/persona.md) | 人设图机制 |
-| [docs/zh/api.md](docs/zh/api.md) | 后端 API 接口 |
-| [docs/zh/prompts.md](docs/zh/prompts.md) | 提示词管理模块 |
-
---
-
-## 特别鸣谢
-
- [Prof. Meiting Wang](https://www.xxmu.edu.cn/yxgcxy/info/1260/4252.htm) — 学术指导
- [逝水秋生白](https://atomgit.com/cenber) — 架构支持
- anzhitinglan — 测试资源支持
- 崔莉萍老师 — 理论指导
- Annie — 专业指导
- 王梓沣、马悦华、隆梦婷 — 神经科学理论支持
-
---
-
-## 许可证
-
-GNU General Public License v3.0 (GPLv3)
-# CI trigger Wednesday, May 13, 2026 AM06:17:55 UTC
-# runner cache cleared Wednesday, May 13, 2026 AM06:23:27 UTC
--- a/README_EN.md
+++ b/README_EN.md
@ -1,79 +0,0 @@
-# TrulyMEM - TrueHumanMEM
-
-<p align="center">
-  <img src="pic/image.png" alt="TrulyMEM Logo" width="200">
-</p>
-
-> **📜 License**: [GNU General Public License v3.0 (GPLv3)](https://www.gnu.org/licenses/gpl-3.0)
-
-> **中文**: [README.md](./README.md)
-
-**Give AI self-awareness, plasticity, and a sense of proportion in long-term memory** — *The More Human Choice.*
-
-[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0)
-[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
-
---
-
-## In a Nutshell
-
-TrulyMEM gives memory authority back to the LLM. Using graph database (triplets) instead of the traditional messages array, the LLM autonomously decides what to remember and what to forget.
-
---
-
-## Screen Shot
-![screenshot_20260508_160449_com.haitai.htbrowser.png](https://raw.gitcode.com/user-images/assets/9658268/9eb36652-4a3b-496a-b957-03b61bac6299/screenshot_20260508_160449_com.haitai.htbrowser.png 'screenshot_20260508_160449_com.haitai.htbrowser.png')
-![screenshot_20260508_160501_com.haitai.htbrowser.png](https://raw.gitcode.com/user-images/assets/9658268/17d96f15-2fab-4647-b272-088f98effeec/screenshot_20260508_160501_com.haitai.htbrowser.png 'screenshot_20260508_160501_com.haitai.htbrowser.png')
-
-## Quick Start
-
-```bash
-python trulymem_entry.py    # from source
-./dist/TrulyMEM             # packaged binary
-```
-
-First run → TUI login screen → create/sign in → press **F2** for API Key → start chatting
-
-📖 **Full guide**: [docs/en/quick_start.md](docs/en/quick_start.md)
-
---
-
-## Features
-
-| Feature | Description |
-|---------|-------------|
-| 🧠 **Graph Memory** | Triplet storage, LLM autonomous navigation |
-| 🔐 **Multi-User** | Isolated profiles + Admin/User roles |
-| 🌐 **Web UI** | Embedded Flask server (thread mode), graph browsing + file upload in chat (PDF/Word/text) |
-| 🎮 **TUI** | Textual-based terminal UI with F2 config panel |
-| 📦 **Single Binary** | PyInstaller build, Web server embedded |
-
---
-
-## Documentation
-
-| Document | Content |
-|----------|---------|
-| [docs/en/quick_start.md](docs/en/quick_start.md) | 🔥 **Full setup guide** (Web, multi-user, building) |
-| [docs/en/architecture.md](docs/en/architecture.md) | System architecture and design |
-| [docs/en/memory.md](docs/en/memory.md) | Memory working mechanism |
-| [docs/en/persona.md](docs/en/persona.md) | Persona Graph mechanism |
-| [docs/en/api.md](docs/en/api.md) | Backend API |
-| [docs/en/prompts.md](docs/en/prompts.md) | Prompt management |
-
---
-
-## Special Thanks
-
- [Prof. Meiting Wang](https://www.xxmu.edu.cn/yxgcxy/info/1260/4252.htm) — Academic guidance
- [逝水秋生白](https://atomgit.com/cenber) — Architecture support
- anzhitinglan — Testing resource support
- 崔莉萍老师 — Theoretical guidance
- Annie — Professional guidance
- 王梓沣、马悦华、隆梦婷 — Neuroscience theory support
-
---
-
-## License
-
-GNU General Public License v3.0 (GPLv3)
--- a/TrulyMEM.spec
+++ b/TrulyMEM.spec
@ -1,80 +1,19 @@
 # -*- mode: python ; coding: utf-8 -*-
-import os
-import sys
+from PyInstaller.utils.hooks import collect_all

-block_cipher = None
+datas = [('ui/styles', 'ui/styles'), ('core/prompts/templates', 'core/prompts/templates'), ('static', 'static'), ('templates', 'templates'), ('core/web_api.py', 'core/')]
+binaries = []
+hiddenimports = ['textual', 'textual.app', 'textual.widgets', 'textual.css', 'openai', 'openai._client', 'neo4j', 'sqlite3', 'core', 'core.embedded_db', 'core.graph_client', 'core.tool_executor', 'core.tool_limiter', 'core.tools', 'core.tools.memory_tools', 'core.prompts', 'core.prompts.prompt_manager', 'core.server', 'core.client', 'core.migrate', 'core.activity_recorder', 'ui', 'ui.app', 'ui.login_screen', 'ui.models', 'ui.models.message', 'ui.models.config', 'ui.models.log_entry', 'ui.widgets', 'ui.handlers', 'ui.services', 'ui.services.config_manager', 'ui.services.config_service', 'core.web_api', 'flask', 'flask_cors', 'werkzeug']
+tmp_ret = collect_all('textual')
+datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]

-project_root = os.path.dirname(os.path.abspath(SPEC))
-sys.path.insert(0, project_root)

-datas = []
-# UI 样式
-ui_styles_dir = os.path.join(project_root, 'ui', 'styles')
-if os.path.exists(ui_styles_dir):
-    for root, dirs, files in os.walk(ui_styles_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'ui/styles'))
-
-# Prompt 模板
-prompt_tmpl_dir = os.path.join(project_root, 'core', 'prompts', 'templates')
-if os.path.exists(prompt_tmpl_dir):
-    for root, dirs, files in os.walk(prompt_tmpl_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'core/prompts/templates'))
-
-# Web 静态文件
-static_dir = os.path.join(project_root, 'ui', 'static')
-if os.path.exists(static_dir):
-    for root, dirs, files in os.walk(static_dir):
-        for f in files:
-            rel_dir = os.path.relpath(root, project_root)
-            datas.append((os.path.join(root, f), rel_dir))
-
-# 兼容旧的 static 目录（如果存在）
-static_dir_old = os.path.join(project_root, 'static')
-if os.path.exists(static_dir_old):
-    for root, dirs, files in os.walk(static_dir_old):
-        for f in files:
-            datas.append((os.path.join(root, f), 'static'))
-
-# Web 模板（Flask template_folder 指向 ui/templates/）
-templates_dir = os.path.join(project_root, 'ui', 'templates')
-if os.path.exists(templates_dir):
-    for root, dirs, files in os.walk(templates_dir):
-        for f in files:
-            rel_dir = os.path.relpath(root, project_root)
-            datas.append((os.path.join(root, f), rel_dir))
-
-# Web API 脚本（以便子进程模式回退使用）
-web_api_src = os.path.join(project_root, 'core', 'web_api.py')
-if os.path.exists(web_api_src):
-    datas.append((web_api_src, 'core'))
-
-# ——— TUI 主二进制 ———
 a = Analysis(
-    [os.path.join(project_root, 'trulymem_entry.py')],
+    ['trulymem_entry.py'],
    pathex=[],
-    binaries=[],
+    binaries=binaries,
    datas=datas,
-    hiddenimports=[
-        'textual', 'textual.app', 'textual.widgets', 'textual.css',
-        'openai', 'openai._client',
-        'neo4j',
-        'sqlite3',
-        'core', 'core.embedded_db', 'core.graph_client',
-        'core.tool_executor', 'core.tool_limiter',
-        'core.tools', 'core.tools.memory_tools',
-        'core.prompts', 'core.prompts.prompt_manager',
-        'core.server', 'core.client', 'core.web_api',
-        'core.migrate',
-        'ui', 'ui.app', 'ui.login_screen',
-        'ui.models', 'ui.models.message', 'ui.models.config', 'ui.models.log_entry',
-        'ui.widgets', 'ui.widgets.left_panel', 'ui.widgets.right_panel',
-        'ui.widgets.input_box', 'ui.widgets.message_history', 'ui.widgets.status_bar',
-        'ui.handlers',
-        'ui.services', 'ui.services.config_manager', 'ui.services.config_service',
-        'flask', 'flask_cors', 'werkzeug',
-    ],
+    hiddenimports=hiddenimports,
    hookspath=[],
    hooksconfig={},
    runtime_hooks=[],
--- a/build-profile.json5
+++ b/build-profile.json5
@ -0,0 +1,91 @@
+{
+  "app": {
+    "signingConfigs": [],
+    "products": [
+      {
+        "name": "default",
+        "targetSdkVersion": "6.1.0(23)",
+        "compatibleSdkVersion": "6.1.0(23)",
+        "runtimeOS": "HarmonyOS",
+        "buildOption": {
+          "strictMode": {
+            "caseSensitiveCheck": true,
+            "useNormalizedOHMUrl": true
+          }
+        }
+      }
+    ],
+    "buildModeSet": [
+      {
+        "name": "debug"
+      },
+      {
+        "name": "release"
+      }
+    ]
+  },
+  "modules": [
+    {
+      "name": "common",
+      "srcPath": "./common",
+      "targets": [
+        {
+          "name": "default",
+          "applyToProducts": [
+            "default"
+          ]
+        }
+      ]
+    },
+
+    {
+      "name": "graph",
+      "srcPath": "./features/graph",
+      "targets": [
+        {
+          "name": "default",
+          "applyToProducts": [
+            "default"
+          ]
+        }
+      ]
+    },
+    {
+      "name": "chat",
+      "srcPath": "./features/chat",
+      "targets": [
+        {
+          "name": "default",
+          "applyToProducts": [
+            "default"
+          ]
+        }
+      ]
+    },
+    {
+      "name": "settings",
+      "srcPath": "./features/settings",
+      "targets": [
+        {
+          "name": "default",
+          "applyToProducts": [
+            "default"
+          ]
+        }
+      ]
+    },
+    {
+      "name": "phone",
+      "srcPath": "./products/phone",
+      "targets": [
+        {
+          "name": "default",
+          "applyToProducts": [
+            "default"
+          ]
+        }
+      ]
+    },
+
+  ]
+}
--- a/build.log
+++ b/build.log
@ -0,0 +1,135 @@
+===== Building TrulyMEM for Linux =====
+Project root: /home/program/TrulyMEM-TrueHumanMEM
+The virtual environment was not created successfully because ensurepip is not
+available.  On Debian/Ubuntu systems, you need to install the python3-venv
+package using the following command.
+
+    apt install python3.13-venv
+
+You may need to use sudo with that command.  After installing the python3-venv
+package, recreate your virtual environment.
+
+Failing command: /home/program/TrulyMEM-TrueHumanMEM/.venv_build/bin/python3
+
+Warning: venv creation failed, falling back to system Python
+Cleaning previous builds...
+================================
+Building TrulyMEM (TUI + Web embedded)
+================================
+29 INFO: PyInstaller: 6.20.0, contrib hooks: 2026.4
+29 INFO: Python: 3.13.12
+31 INFO: Platform: Linux-6.1.0-44-amd64-x86_64-with-glibc2.42
+31 INFO: Python environment: /usr
+33 INFO: Removing temporary files and cleaning cache in /root/.cache/pyinstaller
+34 INFO: Module search paths (PYTHONPATH):
+['/home/program/TrulyMEM-TrueHumanMEM',
+ '/home/program/TrulyMEM-TrueHumanMEM',
+ '/usr/lib/python313.zip',
+ '/usr/lib/python3.13',
+ '/usr/lib/python3.13/lib-dynload',
+ '/usr/local/lib/python3.13/dist-packages',
+ '/usr/lib/python3/dist-packages',
+ '/home/program/TrulyMEM-TrueHumanMEM']
+141 INFO: Appending 'datas' from .spec
+141 INFO: checking Analysis
+141 INFO: Building Analysis because Analysis-00.toc is non existent
+141 INFO: Looking for Python shared library...
+149 INFO: Using Python shared library: /usr/lib/x86_64-linux-gnu/libpython3.13.so.1.0
+149 INFO: Running Analysis Analysis-00.toc
+149 INFO: Target bytecode optimization level: 0
+149 INFO: Initializing module dependency graph...
+149 INFO: Initializing module graph hook caches...
+153 INFO: Analyzing modules for base_library.zip ...
+645 INFO: Processing standard module hook 'hook-encodings.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+1604 INFO: Processing standard module hook 'hook-pickle.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2203 INFO: Processing standard module hook 'hook-heapq.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2434 INFO: Caching module dependency graph...
+2456 INFO: Analyzing /home/program/TrulyMEM-TrueHumanMEM/trulymem_entry.py
+2487 INFO: Processing standard module hook 'hook-sqlite3.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2606 INFO: Processing standard module hook 'hook-platform.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2628 INFO: Processing standard module hook 'hook-sysconfig.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2631 INFO: Processing standard module hook 'hook-_ctypes.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+2641 INFO: Processing pre-safe-import-module hook 'hook-typing_extensions.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+2641 INFO: SetuptoolsInfo: initializing cached setuptools info...
+4602 INFO: Processing standard module hook 'hook-multiprocessing.util.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+4748 INFO: Processing standard module hook 'hook-xml.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+5155 INFO: Processing standard module hook 'hook-pydantic.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+5444 INFO: Processing standard module hook 'hook-rich.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+5729 INFO: Processing standard module hook 'hook-pygments.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+6155 INFO: Processing standard module hook 'hook-chardet.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+7547 INFO: Processing standard module hook 'hook-zoneinfo.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+8697 INFO: Processing standard module hook 'hook-certifi.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+8766 INFO: Processing standard module hook 'hook-anyio.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+9419 INFO: Processing standard module hook 'hook-difflib.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+10529 INFO: Processing standard module hook 'hook-numpy.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+11824 INFO: Processing standard module hook 'hook-charset_normalizer.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+12622 INFO: Processing standard module hook 'hook-pytz.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+13192 INFO: Processing pre-safe-import-module hook 'hook-platformdirs.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+13199 INFO: Processing standard module hook 'hook-platformdirs.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+14917 INFO: Processing standard module hook 'hook-jinja2.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+15284 INFO: Processing pre-safe-import-module hook 'hook-importlib_metadata.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15285 INFO: Setuptools: 'importlib_metadata' appears to be a setuptools-vendored copy - creating alias to 'setuptools._vendor.importlib_metadata'!
+15289 INFO: Processing standard module hook 'hook-setuptools.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+15296 INFO: Processing pre-safe-import-module hook 'hook-distutils.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15314 INFO: Processing pre-safe-import-module hook 'hook-jaraco.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15314 INFO: Setuptools: 'jaraco' appears to be a partial setuptools-vendored copy - extending search paths to ['/usr/lib/python3/dist-packages/jaraco', '/usr/lib/python3/dist-packages/setuptools/_vendor/jaraco']!
+15315 INFO: Processing pre-safe-import-module hook 'hook-jaraco.functools.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15320 INFO: Processing pre-safe-import-module hook 'hook-more_itertools.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15459 INFO: Processing pre-safe-import-module hook 'hook-packaging.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15575 INFO: Processing pre-safe-import-module hook 'hook-jaraco.text.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15581 INFO: Processing standard module hook 'hook-jaraco.text.py' from '/usr/local/lib/python3.13/dist-packages/_pyinstaller_hooks_contrib/stdhooks'
+15614 INFO: Processing pre-safe-import-module hook 'hook-importlib_resources.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15615 INFO: Processing pre-safe-import-module hook 'hook-jaraco.context.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15620 INFO: Processing pre-safe-import-module hook 'hook-backports.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15620 INFO: Setuptools: 'backports' appears to be a full setuptools-vendored copy - creating alias to 'setuptools._vendor.backports'!
+15843 INFO: Processing pre-safe-import-module hook 'hook-tomli.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+15843 INFO: Setuptools: 'tomli' appears to be a setuptools-vendored copy - creating alias to 'setuptools._vendor.tomli'!
+16138 INFO: Processing standard module hook 'hook-pkg_resources.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+16316 INFO: Processing pre-safe-import-module hook 'hook-wheel.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+16376 INFO: Processing standard module hook 'hook-setuptools._vendor.importlib_metadata.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks'
+16378 INFO: Processing pre-safe-import-module hook 'hook-zipp.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/pre_safe_import_module'
+16414 INFO: Analyzing hidden import 'ui.handlers'
+16414 INFO: Analyzing hidden import 'ui.services'
+16415 INFO: Analyzing hidden import 'ui.services.config_manager'
+16416 INFO: Analyzing hidden import 'ui.services.config_service'
+16417 INFO: Processing module hooks (post-graph stage)...
+16716 WARNING: Hidden import "charset_normalizer.md__mypyc" not found!
+18203 INFO: Performing binary vs. data reclassification (622 entries)
+18209 INFO: Looking for ctypes DLLs
+18283 WARNING: Library shell32 required via ctypes not found
+18292 WARNING: Library ole32 required via ctypes not found
+18321 INFO: Analyzing run-time hooks ...
+18329 INFO: Including run-time hook 'pyi_rth_inspect.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/rthooks'
+18331 INFO: Including run-time hook 'pyi_rth_multiprocessing.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/rthooks'
+18333 INFO: Including run-time hook 'pyi_rth_pkgutil.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/rthooks'
+18334 INFO: Including run-time hook 'pyi_rth_setuptools.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/rthooks'
+18335 INFO: Including run-time hook 'pyi_rth_pkgres.py' from '/usr/local/lib/python3.13/dist-packages/PyInstaller/hooks/rthooks'
+18371 INFO: Creating base_library.zip...
+18384 INFO: Looking for dynamic libraries
+18673 INFO: Warnings written to /home/program/TrulyMEM-TrueHumanMEM/build/trulymem/warn-trulymem.txt
+18783 INFO: Graph cross-reference written to /home/program/TrulyMEM-TrueHumanMEM/build/trulymem/xref-trulymem.html
+18816 INFO: checking PYZ
+18816 INFO: Building PYZ because PYZ-00.toc is non existent
+18816 INFO: Building PYZ (ZlibArchive) /home/program/TrulyMEM-TrueHumanMEM/build/trulymem/PYZ-00.pyz
+19768 INFO: Building PYZ (ZlibArchive) /home/program/TrulyMEM-TrueHumanMEM/build/trulymem/PYZ-00.pyz completed successfully.
+19790 WARNING: Ignoring icon; supported only on Windows and macOS!
+19801 INFO: checking PKG
+19801 INFO: Building PKG because PKG-00.toc is non existent
+19801 INFO: Building PKG (CArchive) TrulyMEM.pkg
+24763 INFO: Building PKG (CArchive) TrulyMEM.pkg completed successfully.
+24768 INFO: Bootloader /usr/local/lib/python3.13/dist-packages/PyInstaller/bootloader/Linux-64bit-intel/run
+24768 INFO: checking EXE
+24768 INFO: Building EXE because EXE-00.toc is non existent
+24768 INFO: Building EXE from EXE-00.toc
+24768 INFO: Copying bootloader EXE to /home/program/TrulyMEM-TrueHumanMEM/dist/TrulyMEM
+24768 INFO: Appending PKG archive to custom ELF section in EXE
+24825 INFO: Building EXE from EXE-00.toc completed successfully.
+24830 INFO: Build complete! The results are available in: /home/program/TrulyMEM-TrueHumanMEM/dist
+================================
+===== Build Complete =====
+Binary: dist/TrulyMEM
+total 35848
+drwxr-xr-x  2 root root     4096 Apr 30 07:06 .
+drwxr-xr-x 15 root root     4096 Apr 30 07:05 ..
+-rwxr-xr-x  1 root root 36698096 Apr 30 07:06 TrulyMEM
+Build finished successfully!
--- a/build/build_appimage.sh
+++ b/build/build_appimage.sh
@ -1,100 +0,0 @@
-#!/bin/bash
-set -euo pipefail
-
-echo "===== Building TrulyMEM AppImage ====="
-
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
-PROJECT_ROOT="$(dirname "$SCRIPT_DIR")"
-cd "$PROJECT_ROOT"
-echo "Project root: $PROJECT_ROOT"
-
-APP_NAME="TrulyMEM"
-APP_DIR="$PROJECT_ROOT/build/appimage-build"
-
-# ── Step 1: 复用 build_linux.sh 完成 PyInstaller 构建 ──
-echo ""
-echo "Step 1: Running build_linux.sh (PyInstaller build)..."
-bash "$SCRIPT_DIR/build_linux.sh"
-
-# ── Step 2: 检查 dist/ 中是否有二进制 ──
-echo ""
-echo "Step 2: Checking PyInstaller output..."
-if [ ! -f "dist/$APP_NAME" ]; then
-    echo "Error: dist/$APP_NAME not found after build_linux.sh"; exit 1
-fi
-echo "✅ Found dist/$APP_NAME ($(ls -lh "dist/$APP_NAME" | awk '{print $5}'))"
-
-# ── Step 3: 组织 AppDir 结构 ──
-echo ""
-echo "Step 3: Preparing AppDir structure..."
-rm -rf "$APP_DIR" 2>/dev/null || true
-mkdir -p "$APP_DIR/usr/bin"
-mkdir -p "$APP_DIR/usr/share/applications"
-mkdir -p "$APP_DIR/usr/share/icons/hicolor/256x256/apps"
-mkdir -p "$APP_DIR/usr/share/icons/hicolor/48x48/apps"
-
-cp "dist/$APP_NAME" "$APP_DIR/usr/bin/"
-
-# 图标处理
-ICON_SOURCE=""
-if [ -f "pic/image.png" ]; then
-    ICON_SOURCE="pic/image.png"
-elif [ -f "pic/TrulyMEM.ico" ]; then
-    echo "⚠️  No pic/image.png found; .ico will not display as AppImage icon"
-    echo "   To generate a PNG: convert pic/TrulyMEM.ico pic/image.png"
-fi
-
-if [ -n "$ICON_SOURCE" ]; then
-    cp "$ICON_SOURCE" "$APP_DIR/usr/share/icons/hicolor/256x256/apps/${APP_NAME}.png"
-    cp "$ICON_SOURCE" "$APP_DIR/usr/share/icons/hicolor/48x48/apps/${APP_NAME}.png"
-    cp "$ICON_SOURCE" "$APP_DIR/${APP_NAME}.png"
-    echo "✅ Icon: $ICON_SOURCE"
-else
-    echo "⚠️  No icon found, creating placeholder"
-    touch "$APP_DIR/${APP_NAME}.png"
-fi
-
-# .desktop 文件
-cat > "$APP_DIR/${APP_NAME}.desktop" <<EOF
-[Desktop Entry]
-Name=${APP_NAME}
-Comment=True Human Memory - TUI & Web Mode
-Exec=${APP_NAME}
-Icon=${APP_NAME}
-Type=Application
-Categories=Utility;Office;
-Terminal=true
-StartupNotify=true
-EOF
-
-cp "$APP_DIR/${APP_NAME}.desktop" "$APP_DIR/usr/share/applications/"
-
-# AppRun 入口
-cat > "$APP_DIR/AppRun" <<'APPRUN'
-#!/bin/bash
-HERE="$(dirname "$(readlink -f "$0")")"
-exec "$HERE/usr/bin/TrulyMEM" "$@"
-APPRUN
-chmod +x "$APP_DIR/AppRun"
-
-# ── Step 4: 打包 AppImage ──
-echo ""
-echo "Step 4: Building AppImage..."
-if command -v appimagetool &> /dev/null; then
-    ARCH="${ARCH:-$(uname -m)}" appimagetool "$APP_DIR" "dist/${APP_NAME}.AppImage"
-    echo "✅ AppImage: dist/${APP_NAME}.AppImage"
-    ls -lh "dist/${APP_NAME}.AppImage"
-else
-    echo "⚠️  appimagetool not found. AppDir ready at: $APP_DIR"
-    echo ""
-    echo "To complete manually, install appimagetool:"
-    echo "  wget https://github.com/AppImage/AppImageKit/releases/download/continuous/appimagetool-$(uname -m).AppImage"
-    echo "  chmod +x appimagetool-*.AppImage"
-    echo "  ./appimagetool-*.AppImage '$APP_DIR' 'dist/${APP_NAME}.AppImage'"
-    echo ""
-    echo "AppDir contents:"
-    find "$APP_DIR" -type f | head -20
-fi
-
-echo ""
-echo "===== AppImage Build Complete ====="
--- a/build/build_linux.sh
+++ b/build/build_linux.sh
@ -1,109 +0,0 @@
-#!/bin/bash
-set -euo pipefail
-
-echo "===== Building TrulyMEM for Linux ====="
-
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
-PROJECT_ROOT="$(dirname "$SCRIPT_DIR")"
-cd "$PROJECT_ROOT"
-echo "Project root: $PROJECT_ROOT"
-
-# ── 前置检查 ──
-if ! command -v python3 &> /dev/null; then
-    echo "Error: python3 not found"; exit 1
-fi
-
-if [ ! -f requirements.txt ]; then
-    echo "Error: requirements.txt not found in $PROJECT_ROOT"; exit 1
-fi
-
-# ── 检测 python3-venv ──
-VENV_AVAILABLE=false
-if python3 -c "import ensurepip" 2>/dev/null && python3 -m venv --help &>/dev/null; then
-    VENV_AVAILABLE=true
-else
-    echo "⚠️  python3-venv 未安装（或缺少 ensurepip），建议安装以获得干净构建环境："
-    echo "   sudo apt install python3-venv    # Debian/Ubuntu"
-    echo "   sudo dnf install python3-virtualenv  # Fedora"
-    echo "将使用系统 Python 环境继续（依赖全局包）..."
-    echo ""
-fi
-
-# ── 尝试 venv 隔离构建，失败则用系统环境 ──
-USE_VENV=false
-if [ "$VENV_AVAILABLE" = true ]; then
-    VENV_DIR="$PROJECT_ROOT/.venv_build"
-    echo "Creating virtual environment..."
-    if python3 -m venv "$VENV_DIR" 2>/dev/null; then
-        source "$VENV_DIR/bin/activate"
-        USE_VENV=true
-        echo "✅ Using virtual environment: $VENV_DIR"
-        pip install --upgrade pip -q
-        pip install -r requirements.txt -q
-    else
-        echo "⚠️  venv creation failed, falling back to system Python"
-    fi
-fi
-
-if [ "$USE_VENV" = false ]; then
-    echo "Installing dependencies (system Python)..."
-    pip install -r requirements.txt --break-system-packages -q 2>/dev/null || \
-        pip install -r requirements.txt -q 2>/dev/null || {
-            echo "⚠️  pip install failed, trying pip3..."
-            pip3 install -r requirements.txt --break-system-packages -q 2>/dev/null || \
-            pip3 install -r requirements.txt -q 2>/dev/null || \
-            echo "⚠️  Some dependencies may be missing; build will proceed anyway"
-        }
-fi
-
-# ── 清理旧构建 ──
-echo ""
-echo "Cleaning previous builds..."
-rm -rf dist/ build/trulymem/ 2>/dev/null || true
-
-# ── PyInstaller 构建 ──
-echo ""
-echo "================================"
-echo "Building TrulyMEM (TUI + Web embedded)"
-echo "================================"
-python3 -m PyInstaller --clean build/trulymem.spec --noconfirm
-
-# ── 创建 Linux `.desktop` 文件（打包图标不能嵌入 ELF，通过 .desktop 引用）──
-echo ""
-echo "Generating .desktop file for Linux..."
-BINARY_PATH="$(cd dist && pwd)/TrulyMEM"
-ICON_PATH="$(cd pic && pwd)/image.png"
-
-cat > "dist/TrulyMEM.desktop" <<EOF
-[Desktop Entry]
-Name=TrulyMEM
-Comment=True Human Memory - TUI & Web Mode
-Exec=${BINARY_PATH}
-Icon=${ICON_PATH}
-Terminal=true
-Type=Application
-Categories=Utility;Office;
-StartupNotify=true
-EOF
-
-chmod +x "dist/TrulyMEM.desktop"
-echo "✅ dist/TrulyMEM.desktop created (icon: pic/image.png)"
-
-# ── 构建完成 ──
-echo ""
-echo "================================"
-echo "===== Build Complete ====="
-echo "Binary:     dist/TrulyMEM"
-echo "Desktop:    dist/TrulyMEM.desktop"
-echo "------------------------------"
-ls -lh dist/ 2>/dev/null || ls -la dist/
-echo "================================"
-
-# ── 清理 venv ──
-if [ "$USE_VENV" = true ]; then
-    deactivate 2>/dev/null || true
-    rm -rf "$VENV_DIR"
-    echo "Virtual environment cleaned up."
-fi
-
-echo "Build finished successfully!"
--- a/build/build_macos.sh
+++ b/build/build_macos.sh
@ -1,38 +0,0 @@
-#!/bin/bash
-set -e
-
-echo "===== Building TrulyMEM for macOS ====="
-
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
-PROJECT_ROOT="$(dirname "$SCRIPT_DIR")"
-cd "$PROJECT_ROOT"
-echo "Project root: $PROJECT_ROOT"
-
-if ! command -v python3 &> /dev/null; then
-    echo "Error: python3 not found"; exit 1
-fi
-
-VENV_DIR="$PROJECT_ROOT/.venv_build"
-echo "Creating virtual environment: $VENV_DIR"
-python3 -m venv "$VENV_DIR"
-source "$VENV_DIR/bin/activate"
-pip install --upgrade pip
-pip install -r requirements.txt
-
-echo "Cleaning previous builds..."
-rm -rf dist/ build/trulymem/ 2>/dev/null || true
-
-echo "================================"
-echo "Building TrulyMEM (TUI + Web embedded)"
-echo "================================"
-python -m PyInstaller --clean build/trulymem.spec --noconfirm
-
-echo "================================"
-echo "===== Build Complete ====="
-echo "Binary: dist/TrulyMEM"
-echo "  -> run: open dist/TrulyMEM"
-ls -la dist/
-
-deactivate
-rm -rf "$VENV_DIR"
-echo "Build finished successfully!"
--- a/build/build_windows.bat
+++ b/build/build_windows.bat
@ -1,35 +0,0 @@
-@echo off
-setlocal enabledelayedexpansion
-
-echo ===== Building TrulyMEM for Windows =====
-
-set "SCRIPT_DIR=%~dp0"
-set "PROJECT_ROOT=%SCRIPT_DIR%.."
-cd /d "%PROJECT_ROOT%"
-echo Project root: %PROJECT_ROOT%
-
-echo Creating virtual environment: .venv_build
-python -m venv .venv_build
-call .venv_build\Scripts\activate.bat
-pip install --upgrade pip
-pip install -r requirements.txt
-
-echo Cleaning previous builds...
-if exist dist rmdir /s /q dist
-if exist build\trulymem rmdir /s /q build\trulymem
-
-echo ================================
-echo Building TrulyMEM (TUI + Web embedded)
-echo ================================
-python -m PyInstaller --clean build\trulymem.spec --noconfirm
-
-echo ================================
-echo ===== Build Complete =====
-echo Binary: dist\TrulyMEM.exe
-dir dist
-
-call .venv_build\Scripts\deactivate.bat
-if exist .venv_build rmdir /s /q .venv_build
-
-echo Build finished successfully!
-endlocal
--- a/build/trulymem.spec
+++ b/build/trulymem.spec
@ -1,99 +0,0 @@
-# -*- mode: python ; coding: utf-8 -*-
-import os
-import sys
-
-block_cipher = None
-
-project_root = os.path.dirname(os.path.dirname(os.path.abspath(SPEC)))
-sys.path.insert(0, project_root)
-
-datas = []
-# UI 样式
-ui_styles_dir = os.path.join(project_root, 'ui', 'styles')
-if os.path.exists(ui_styles_dir):
-    for root, dirs, files in os.walk(ui_styles_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'ui/styles'))
-
-# Prompt 模板
-prompt_tmpl_dir = os.path.join(project_root, 'core', 'prompts', 'templates')
-if os.path.exists(prompt_tmpl_dir):
-    for root, dirs, files in os.walk(prompt_tmpl_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'core/prompts/templates'))
-
-# Web 静态文件
-static_dir = os.path.join(project_root, 'ui', 'static')
-if os.path.exists(static_dir):
-    for root, dirs, files in os.walk(static_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'ui/static'))
-
-# Web 模板
-templates_dir = os.path.join(project_root, 'ui', 'templates')
-if os.path.exists(templates_dir):
-    for root, dirs, files in os.walk(templates_dir):
-        for f in files:
-            datas.append((os.path.join(root, f), 'ui/templates'))
-
-# Web API 脚本（以便子进程模式回退使用）
-web_api_src = os.path.join(project_root, 'core', 'web_api.py')
-if os.path.exists(web_api_src):
-    datas.append((web_api_src, '.'))
-
-# ——— TUI 主二进制 ———
-a = Analysis(
-    [os.path.join(project_root, 'trulymem_entry.py')],
-    pathex=[],
-    binaries=[],
-    datas=datas,
-    hiddenimports=[
-        'textual', 'textual.app', 'textual.widgets', 'textual.css',
-        'openai', 'openai._client',
-        'neo4j',
-        'sqlite3',
-        'core', 'core.embedded_db', 'core.graph_client',
-        'core.tool_executor', 'core.tool_limiter',
-        'core.tools', 'core.tools.memory_tools',
-        'core.prompts', 'core.prompts.prompt_manager',
-        'core.server', 'core.client',
-        'core.migrate',
-        'ui', 'ui.app', 'ui.login_screen',
-        'ui.models', 'ui.models.message', 'ui.models.config', 'ui.models.log_entry',
-        'ui.widgets', 'ui.widgets.left_panel', 'ui.widgets.right_panel',
-        'ui.widgets.input_box', 'ui.widgets.message_history', 'ui.widgets.status_bar',
-        'ui.handlers',
-        'ui.services', 'ui.services.config_manager', 'ui.services.config_service',
-        'web_api',
-        'flask', 'flask_cors', 'werkzeug',
-    ],
-    hookspath=[],
-    hooksconfig={},
-    runtime_hooks=[],
-    excludes=[],
-    noarchive=False,
-    optimize=0,
-)
-pyz = PYZ(a.pure)
-
-exe = EXE(
-    pyz,
-    a.scripts,
-    a.binaries,
-    a.datas,
-    [],
-    name='TrulyMEM',
-    icon=os.path.join(project_root, 'pic', 'TrulyMEM.ico'),
-    debug=False,
-    bootloader_ignore_signals=False,
-    strip=False,
-    upx=True,
-    upx_exclude=[],
-    runtime_tmpdir=None,
-    console=True,
-    disable_windowed_traceback=False,
-    argv_emulation=False,
-    target_arch=None,
-    codesign_identity=None,
-    entitlements_file=None,
-)
--- a/code-linter.json5
+++ b/code-linter.json5
@ -0,0 +1,32 @@
+{
+  "files": [
+    "**/*.ets"
+  ],
+  "ignore": [
+    "**/src/ohosTest/**/*",
+    "**/src/test/**/*",
+    "**/src/mock/**/*",
+    "**/node_modules/**/*",
+    "**/oh_modules/**/*",
+    "**/build/**/*",
+    "**/.preview/**/*"
+  ],
+  "ruleSet": [
+    "plugin:@performance/recommended",
+    "plugin:@typescript-eslint/recommended"
+  ],
+  "rules": {
+    "@security/no-unsafe-aes": "error",
+    "@security/no-unsafe-hash": "error",
+    "@security/no-unsafe-mac": "warn",
+    "@security/no-unsafe-dh": "error",
+    "@security/no-unsafe-dsa": "error",
+    "@security/no-unsafe-ecdsa": "error",
+    "@security/no-unsafe-rsa-encrypt": "error",
+    "@security/no-unsafe-rsa-sign": "error",
+    "@security/no-unsafe-rsa-key": "error",
+    "@security/no-unsafe-dsa-key": "error",
+    "@security/no-unsafe-dh-key": "error",
+    "@security/no-unsafe-3des": "error"
+  }
+}
--- a/products/phone/build/default/generated/profile/default/BuildProfile.ets
+++ b/products/phone/build/default/generated/profile/default/BuildProfile.ets
@ -1,25 +1,17 @@
 /**
 * Use these variables when you tailor your ArkTS code. They must be of the const type.
 */
-export const BUNDLE_NAME = 'com.trulymem.app';
-export const BUNDLE_TYPE = 'app';
-export const VERSION_CODE = 1000001;
-export const VERSION_NAME = '1.0.0';
-export const TARGET_NAME = 'default';
-export const PRODUCT_NAME = 'default';
+export const HAR_VERSION = '1.0.0';
 export const BUILD_MODE_NAME = 'debug';
 export const DEBUG = true;
+export const TARGET_NAME = 'default';

 /**
 * BuildProfile Class is used only for compatibility purposes.
 */
 export default class BuildProfile { 
-	static readonly BUNDLE_NAME = BUNDLE_NAME;
-	static readonly BUNDLE_TYPE = BUNDLE_TYPE;
-	static readonly VERSION_CODE = VERSION_CODE;
-	static readonly VERSION_NAME = VERSION_NAME;
-	static readonly TARGET_NAME = TARGET_NAME;
-	static readonly PRODUCT_NAME = PRODUCT_NAME;
+	static readonly HAR_VERSION = HAR_VERSION;
 	static readonly BUILD_MODE_NAME = BUILD_MODE_NAME;
 	static readonly DEBUG = DEBUG;
+	static readonly TARGET_NAME = TARGET_NAME;
 }
--- a/common/Index.ets
+++ b/common/Index.ets
@ -0,0 +1,24 @@
+// ========= Utility Layer =========
+export { defaultLogger } from './src/main/ets/util/Logger';
+export { defaultLogger as Logger } from './src/main/ets/util/Logger';
+
+export { BreakpointType, BreakpointTypes, WidthBreakpoint } from "./src/main/ets/util/BreakpointSystem";
+
+// ========= Router =========
+export { PageContext, RouterParam, IPageContext } from "./src/main/ets/routermanager/PageContext";
+
+// ========= Constants =========
+export { Constants as TrulyMEMConstants } from "./src/main/ets/constant/TrulyMEMConstants";
+
+// ========= Model Layer =========
+export { GraphDatabase, RecallEntity, TimeRangeParams } from "./src/main/ets/model/GraphDatabase";
+
+// ========= Service Layer =========
+export { GraphMemoryService, ConnectionItem, NodeDetailInfo } from "./src/main/ets/service/GraphMemoryService";
+export { AIAgentService, ChatMessage, AgentResponse } from "./src/main/ets/service/AIAgentService";
+
+// ========= ViewModel Layer =========
+export { BaseViewModel, VMEvent } from "./src/main/ets/viewmodel/BaseViewModel";
+
+// ========= Component Layer =========
+export { ImmersiveTabNavigation } from "./src/main/ets/component/ImmersiveTabNavigation";
--- a/common/build-profile.json5
+++ b/common/build-profile.json5
@ -0,0 +1,8 @@
+{
+  "apiType": "stageMode",
+  "targets": [
+    {
+      "name": "default"
+    }
+  ]
+}
--- a/common/build/default/intermediates/merge_profile/default/module.json
+++ b/common/build/default/intermediates/merge_profile/default/module.json
@ -1,30 +0,0 @@
-{
-	"app": {
-		"bundleName": "com.trulymem.app",
-		"debug": true,
-		"versionCode": 1000001,
-		"versionName": "1.0.0",
-		"minAPIVersion": 60100023,
-		"targetAPIVersion": 60100023,
-		"apiReleaseType": "Release",
-		"targetMinorAPIVersion": 0,
-		"targetPatchAPIVersion": 0,
-		"compileSdkVersion": "6.1.0.105",
-		"compileSdkType": "HarmonyOS",
-		"appEnvironments": [],
-		"bundleType": "app",
-		"buildMode": "debug"
-	},
-	"module": {
-		"name": "common",
-		"type": "har",
-		"description": "TrulyMEM common module",
-		"deviceTypes": [
-			"phone",
-			"tablet",
-			"2in1"
-		],
-		"packageName": "@ohos/common",
-		"installationFree": false
-	}
-}
--- a/common/common
+++ b/common/common
@ -0,0 +1 @@
+/home/program/TrulyMEM-TrueHumanMEM/common
--- a/common/hvigorfile.ts
+++ b/common/hvigorfile.ts
@ -0,0 +1,6 @@
+import { harTasks } from '@ohos/hvigor-ohos-plugin';
+
+export default {
+    system: harTasks,
+    plugins: []
+};
--- a/common/oh-package.json5
+++ b/common/oh-package.json5
@ -0,0 +1,9 @@
+{
+  "name": "@ohos/common",
+  "version": "1.0.0",
+  "description": "TrulyMEM common module",
+  "main": "Index.ets",
+  "author": "",
+  "license": "",
+  "dependencies": {}
+}
--- a/common/src/main/ets/component/ImmersiveTabNavigation.ets
+++ b/common/src/main/ets/component/ImmersiveTabNavigation.ets
@ -0,0 +1,133 @@
+import { defaultLogger } from '../util/Logger';
+import { window } from '@kit.ArkUI';
+import { BusinessError } from '@kit.BasicServicesKit';
+
+const THEME_COLOR = '#7C4DFF';
+
+@Component
+export struct ImmersiveTabNavigation {
+  @State currentIndex: number = 0;
+  @BuilderParam contentBuilder: () => void;
+  onTabChange?: (index: number) => void;
+
+  private windowFocused: boolean = true;
+  private bottomAvoidHeight: number = 0;
+
+  aboutToAppear() {
+    const mainWindow = AppStorage.get<window.Window>('main_window');
+    if (mainWindow) {
+      try {
+        const avoidArea = mainWindow.getWindowAvoidArea(window.AvoidAreaType.TYPE_SYSTEM);
+        this.bottomAvoidHeight = avoidArea.bottomRect.height || 0;
+      } catch (e) {
+        defaultLogger.error('Failed to get avoid area: ' + (e as BusinessError).message);
+      }
+    }
+  }
+
+  triggerTabSwitchFeedback(index: number) {
+    this.currentIndex = index;
+    AppStorage.setOrCreate('global_theme_color', THEME_COLOR);
+    this.onTabChange?.(index);
+  }
+
+  @Builder
+  tabBarBuilder(index: number, icon: string, label: string) {
+    Column() {
+      if (this.currentIndex === index && this.windowFocused) {
+        Circle()
+          .width(32)
+          .height(32)
+          .backgroundColor(`${THEME_COLOR}33`)
+          .blur(8)
+          .position({ x: '50%', y: '50%' })
+          .translate({ x: '-50%', y: '-50%' })
+      }
+
+      Text(icon)
+        .fontSize(20)
+        .opacity(this.currentIndex === index ? 1 : 0.5)
+
+      Text(label)
+        .fontSize(10)
+        .fontColor(this.currentIndex === index ? THEME_COLOR : '#999')
+        .fontWeight(this.currentIndex === index ? FontWeight.Bold : FontWeight.Normal)
+    }
+    .width('100%')
+    .height(56)
+    .justifyContent(FlexAlign.Center)
+    .alignItems(HorizontalAlign.Center)
+  }
+
+  build() {
+    Stack() {
+      Column() {
+        this.contentBuilder()
+      }
+      .width('100%')
+      .height('100%')
+
+      Column() {
+        Stack() {
+          Column()
+            .width('100%')
+            .height('100%')
+            .backgroundBlurStyle(BlurStyle.Regular)
+            .borderRadius(24)
+
+          Column()
+            .width('100%')
+            .height('100%')
+            .backgroundColor(`${THEME_COLOR}0D`)
+            .borderRadius(24)
+
+          Column()
+            .width('100%')
+            .height('100%')
+            .linearGradient({
+              angle: 180,
+              colors: [['rgba(255,255,255,0.15)', 0.0], ['rgba(255,255,255,0.05)', 1.0]]
+            })
+            .borderRadius(24)
+        }
+        .width('100%')
+        .height('100%')
+
+        Tabs({ index: this.currentIndex }) {
+          TabContent() {
+            Column() {
+              Blank()
+            }
+          }
+          .tabBar(this.tabBarBuilder(0, '🌌', 'TrulyMEM'))
+
+          TabContent() {
+            Column() {
+              Blank()
+            }
+          }
+          .tabBar(this.tabBarBuilder(1, '⚙', '设置'))
+        }
+        .width('100%')
+        .height(64)
+        .barPosition(BarPosition.End)
+        .onChange((index: number) => {
+          this.triggerTabSwitchFeedback(index);
+        })
+      }
+      .width('92%')
+      .height(72)
+      .alignSelf(ItemAlign.Center)
+      .position({ y: `calc(100% - ${this.bottomAvoidHeight > 0 ? this.bottomAvoidHeight : 16}px - 72px)` })
+      .borderRadius(24)
+      .shadow({
+        radius: 20,
+        offsetY: -4,
+        color: 'rgba(0,0,0,0.15)'
+      })
+    }
+    .width('100%')
+    .height('100%')
+    .backgroundColor('#00000000')
+  }
+}
--- a/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/constant/TrulyMEMConstants.ts
+++ b/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/constant/TrulyMEMConstants.ts
@ -4,10 +4,12 @@ export class Constants {
    static readonly DEFAULT_BASE_URL: string = 'https://api.deepseek.com';
    static readonly DEFAULT_MODEL: string = 'deepseek-chat';
    static readonly SECURITY_LEVEL: number = 1; // S1
+
    // Table names
    static readonly TABLE_NODES: string = 'nodes';
    static readonly TABLE_RELATIONS: string = 'relations';
    static readonly TABLE_CHAT: string = 'chat_records';
+
    // SQL definitions
    static readonly SQL_CREATE_NODES: string = `
        CREATE TABLE IF NOT EXISTS nodes (
@ -18,6 +20,7 @@ export class Constants {
            created_at TEXT DEFAULT (datetime('now','localtime')),
            updated_at TEXT DEFAULT (datetime('now','localtime'))
        )`;
+
    static readonly SQL_CREATE_RELATIONS: string = `
        CREATE TABLE IF NOT EXISTS relations (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
@ -29,6 +32,7 @@ export class Constants {
            FOREIGN KEY (subject_id) REFERENCES nodes(id) ON DELETE CASCADE,
            FOREIGN KEY (object_id) REFERENCES nodes(id) ON DELETE CASCADE
        )`;
+
    static readonly SQL_CREATE_CHAT: string = `
        CREATE TABLE IF NOT EXISTS chat_records (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
--- a/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/model/GraphDatabase.ts
+++ b/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/model/GraphDatabase.ts
@ -1,18 +1,15 @@
-import relationalStore from "@ohos:data.relationalStore";
-import type common from "@ohos:app.ability.common";
-interface NodeNameCacheItem {
-    name: string;
-    type: string;
-    mentions: number;
-}
-export interface TimeRangeParams {
-    days: number;
-}
+import relationalStore from '@ohos.data.relationalStore';
+import { Context } from '@ohos.abilityAccessCtrl';
+
+interface NodeNameCacheItem { name: string; type: string; mentions: number; }
+export interface TimeRangeParams { days: number; }
+
 interface TripletData {
    subject: string;
    relation: string;
    object: string;
 }
+
 interface CriteriaData {
    subject?: string;
    target?: string;
@ -24,6 +21,7 @@ interface CriteriaData {
    subjectContains?: string;
    targetContains?: string;
 }
+
 interface NodeData {
    id: number;
    label: string;
@ -31,6 +29,7 @@ interface NodeData {
    mentions: number;
    depth?: number;
 }
+
 interface EdgeData {
    from: number;
    to: number;
@ -40,16 +39,19 @@ interface EdgeData {
    sessionId?: string;
    turnId?: number;
 }
+
 interface GraphData {
    nodes: NodeData[];
    edges: EdgeData[];
 }
+
 export interface RecallEntity {
    name: string;
    type: string;
    mention_count: number;
    depth?: number;
 }
+
 interface RecallRelation {
    source: string;
    target: string;
@ -59,11 +61,13 @@ interface RecallRelation {
    turn_id?: number;
    depth?: number;
 }
+
 interface RecallResult {
    entities: RecallEntity[];
    relations: RecallRelation[];
    message: string;
 }
+
 interface BfsEntity {
    id: number;
    name: string;
@ -71,6 +75,7 @@ interface BfsEntity {
    mentions: number;
    depth: number;
 }
+
 export interface RelationQueryResult {
    sourceId: number;
    targetId: number;
@ -82,6 +87,7 @@ export interface RelationQueryResult {
    turnId?: number;
    depth: number;
 }
+
 interface NodeQueryResult {
    id: number;
    name: string;
@ -89,6 +95,7 @@ interface NodeQueryResult {
    mentions: number;
    depth: number;
 }
+
 interface CleanupResult {
    cleaned: number;
    deleted_relations?: number;
@ -96,25 +103,30 @@ interface CleanupResult {
    dry_run?: boolean;
    message?: string;
 }
+
 interface IntrospectResult {
    entity_count: number;
    relation_count: number;
    message: string;
 }
+
 interface ArchiveResult {
    archived: number;
    message: string;
 }
+
 interface SearchResultItem {
    name: string;
    type: string;
    mentions: number;
 }
+
 class TaskNodeRow {
    id: number = 0;
    name: string = '';
    updatedAt: string = '';
 }
+
 interface DbTaskInfo {
    taskId: string;
    description: string;
@ -122,30 +134,35 @@ interface DbTaskInfo {
    infoCount: number;
    updatedAt: string;
 }
+
 interface ChatMessage {
    role: string;
    content: string;
    session_id?: string;
 }
+
 interface SnapshotData {
    entities: RecallEntity[];
    relations: RecallRelation[];
 }
+
 const STORE_CONFIG: relationalStore.StoreConfig = {
    name: 'trulymem.db',
    securityLevel: relationalStore.SecurityLevel.S1
 };
+
 export class GraphDatabase {
    private store?: relationalStore.RdbStore;
-    private context?: common.Context;
-    async init(context: common.Context): Promise<void> {
+    private context?: Context;
+
+    async init(context: Context): Promise<void> {
        this.context = context;
        this.store = await relationalStore.getRdbStore(context, STORE_CONFIG);
        await this.createTables();
    }
+
    private async createTables(): Promise<void> {
-        if (!this.store)
-            return;
+        if (!this.store) return;
        await this.store.executeSql(`
            CREATE TABLE IF NOT EXISTS nodes (
                id INTEGER PRIMARY KEY AUTOINCREMENT,
@ -188,9 +205,9 @@ export class GraphDatabase {
        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_type ON relations(relation)`);
        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_status ON relations(status)`);
    }
+
    async commit(triplets: TripletData[], entityTypes?: Record<string, string>, sessionId?: string, turnId?: number): Promise<void> {
-        if (!this.store)
-            return;
+        if (!this.store) return;
        for (const triplet of triplets) {
            const subjectId: number = await this.upsertNode(triplet.subject, entityTypes?.[triplet.subject]);
            const objectId: number = await this.upsertNode(triplet.object, entityTypes?.[triplet.object]);
@ -214,9 +231,9 @@ export class GraphDatabase {
            await this.store.insert('relations', bucket);
        }
    }
+
    private async checkDuplicateRelation(subjectId: number, relation: string, objectId: number): Promise<number> {
-        if (!this.store)
-            return -1;
+        if (!this.store) return -1;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        predicates.equalTo('subject_id', subjectId).and().equalTo('relation', relation).and().equalTo('object_id', objectId).and().equalTo('status', 'active');
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id']);
@ -228,9 +245,9 @@ export class GraphDatabase {
        resultSet.close();
        return -1;
    }
+
    private async upsertNode(name: string, entityType?: string): Promise<number> {
-        if (!this.store)
-            return -1;
+        if (!this.store) return -1;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        predicates.equalTo('name', name);
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'mentions']);
@ -258,6 +275,7 @@ export class GraphDatabase {
        };
        return await this.store.insert('nodes', bucket);
    }
+
    async recall(queryIntent: string, seedEntities?: string[], depth: number = 2, timeRange?: TimeRangeParams, sessionFilter?: string): Promise<RecallResult> {
        if (!this.store) {
            return { entities: [], relations: [], message: 'Database not initialized' };
@ -273,6 +291,7 @@ export class GraphDatabase {
        const allEntities: BfsEntity[] = [];
        const entityIds = new Set<number>();
        let seedEntityIds = new Set<number>();
+
        if (keywords.length === 0 && (!seedEntities || seedEntities.length === 0)) {
            const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
            predicates.orderByDesc('mentions');
@ -286,8 +305,7 @@ export class GraphDatabase {
                allEntities.push({ id, name, type, mentions, depth: 0 });
            }
            resultSet.close();
-        }
-        else {
+        } else {
            if (seedEntities && seedEntities.length > 0) {
                for (const seedName of seedEntities) {
                    const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
@ -330,15 +348,18 @@ export class GraphDatabase {
                resultSet.close();
            }
        }
+
        const allRelations: RelationQueryResult[] = [];
        let currentLayerIds = new Set<number>(entityIds);
        const visitedEntityIds = new Set<number>(entityIds);
        // 批量预加载所有相关节点名称，减少 N+1 查询
        const nodeNameCache = new Map<number, NodeNameCacheItem>();
+
        for (let layer = 0; layer < depth && currentLayerIds.size > 0; layer++) {
            const currentIds = Array.from(currentLayerIds);
            const relations = await this.getRelationsForNodes(currentIds, sessionFilter, minDateBucket);
            const nextLayerIds = new Set<number>();
+
            for (const rel of relations) {
                allRelations.push(rel);
                if (!visitedEntityIds.has(rel.targetId)) {
@ -348,6 +369,7 @@ export class GraphDatabase {
                    nextLayerIds.add(rel.sourceId);
                }
            }
+
            for (const newId of nextLayerIds) {
                if (!visitedEntityIds.has(newId)) {
                    visitedEntityIds.add(newId);
@ -362,8 +384,7 @@ export class GraphDatabase {
                            depth: layer + 1
                        };
                        allEntities.push(addedEntity);
-                    }
-                    else {
+                    } else {
                        const nodeData = await this.getNodeById(newId);
                        if (nodeData) {
                            const cacheItem: NodeNameCacheItem = { name: nodeData.name, type: nodeData.type, mentions: nodeData.mentions };
@ -382,6 +403,7 @@ export class GraphDatabase {
            }
            currentLayerIds = nextLayerIds;
        }
+
        const entities: RecallEntity[] = allEntities.map(e => {
            const entity: RecallEntity = {
                name: e.name,
@ -403,15 +425,16 @@ export class GraphDatabase {
            };
            return rel;
        });
+
        return {
            entities,
            relations,
            message: `找到 ${entities.length} 个实体, ${relations.length} 条关系`
        };
    }
+
    private async getNodeById(id: number): Promise<NodeQueryResult | null> {
-        if (!this.store)
-            return null;
+        if (!this.store) return null;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        predicates.equalTo('id', id);
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
@ -429,9 +452,9 @@ export class GraphDatabase {
        resultSet.close();
        return null;
    }
+
    private async getRelationsForNodes(nodeIds: number[], sessionFilter?: string, minDateBucket?: string): Promise<RelationQueryResult[]> {
-        if (!this.store || nodeIds.length === 0)
-            return [];
+        if (!this.store || nodeIds.length === 0) return [];
        const relations: RelationQueryResult[] = [];
        // 批量预加载所有节点名称到缓存，避免 N+1 查询
        const nodeNameCache = new Map<number, NodeNameCacheItem>();
@ -472,6 +495,7 @@ export class GraphDatabase {
                }
            }
            resultSet.close();
+
            const predicates2: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
            predicates2.equalTo('status', 'active').and().equalTo('object_id', nodeId);
            if (sessionFilter) {
@ -504,9 +528,9 @@ export class GraphDatabase {
        }
        return relations;
    }
+
    async search(keyword: string): Promise<SearchResultItem[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        predicates.like('name', `%${keyword}%`);
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['name', 'type', 'mentions']);
@ -521,11 +545,12 @@ export class GraphDatabase {
        resultSet.close();
        return results;
    }
+
    async purge(criteria: CriteriaData, mode: string = 'soft'): Promise<void> {
-        if (!this.store)
-            return;
+        if (!this.store) return;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        let hasCondition = false;
+
        // 精确匹配
        if (criteria.subject) {
            const subjectId = await this.getNodeIdByName(criteria.subject);
@ -558,6 +583,7 @@ export class GraphDatabase {
            predicates.equalTo('session_id', criteria.sessionId);
            hasCondition = true;
        }
+
        // 模糊匹配：subjectContains -> 通过子查询匹配节点名
        if (criteria.subjectContains) {
            const nodeSql = `SELECT id FROM nodes WHERE name LIKE '%' || ? || '%'`;
@ -575,6 +601,7 @@ export class GraphDatabase {
                hasCondition = true;
            }
        }
+
        // 模糊匹配：targetContains
        if (criteria.targetContains) {
            const nodeSql = `SELECT id FROM nodes WHERE name LIKE '%' || ? || '%'`;
@ -592,6 +619,7 @@ export class GraphDatabase {
                hasCondition = true;
            }
        }
+
        // 源实体类型过滤
        if (criteria.sourceType) {
            const nodeSql = `SELECT id FROM nodes WHERE type = ?`;
@ -609,6 +637,7 @@ export class GraphDatabase {
                hasCondition = true;
            }
        }
+
        // 目标实体类型过滤
        if (criteria.targetType) {
            const nodeSql = `SELECT id FROM nodes WHERE type = ?`;
@ -626,6 +655,7 @@ export class GraphDatabase {
                hasCondition = true;
            }
        }
+
        // 源实体状态过滤
        if (criteria.sourceHasStatus) {
            const nodeSql = `SELECT id FROM nodes WHERE type LIKE '%' || ? || '%'`;
@ -643,6 +673,7 @@ export class GraphDatabase {
                hasCondition = true;
            }
        }
+
        if (hasCondition) {
            if (mode === 'soft') {
                const bucket: relationalStore.ValuesBucket = {
@ -650,20 +681,19 @@ export class GraphDatabase {
                    'updated_at': new Date().toISOString()
                };
                await this.store.update(bucket, predicates);
-            }
-            else {
+            } else {
                await this.store.delete(predicates);
            }
        }
        await this.removeOrphanNodes();
    }
+
    /**
     * 记忆图谱 — 在指定时间范围内查询关系和节点
     * 对应 tools.memory_graph
     */
    async graph(timeRange: TimeRangeParams, sessionFilter?: string): Promise<GraphData> {
-        if (!this.store)
-            return { nodes: [], edges: [] };
+        if (!this.store) return { nodes: [], edges: [] };
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        predicates.equalTo('status', 'active');
        if (sessionFilter) {
@ -705,13 +735,13 @@ export class GraphDatabase {
        const result: GraphData = { nodes, edges };
        return result;
    }
+
    /**
     * 记忆快照 — 在指定时间范围内查询实体和关系
     * 对应 tools.memory_snapshot
     */
    async snapshot(timeRange: TimeRangeParams, sessionFilter?: string): Promise<SnapshotData> {
-        if (!this.store)
-            return { entities: [], relations: [] };
+        if (!this.store) return { entities: [], relations: [] };
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        predicates.equalTo('status', 'active');
        if (sessionFilter) {
@ -757,13 +787,13 @@ export class GraphDatabase {
        const snapResult: SnapshotData = { entities, relations };
        return snapResult;
    }
+
    /**
     * 查询已归档的记忆
     * 对应 tools.memory_query_archived
     */
    async queryArchived(days?: number, keyword?: string): Promise<RelationQueryResult[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        let sql = `
            SELECT r.subject_id, r.object_id, r.relation, r.weight, r.session_id, r.turn_id,
                   e.name AS source_name, t.name AS target_name
@ -804,9 +834,9 @@ export class GraphDatabase {
        resultSet.close();
        return results;
    }
+
    async getRecentTasks(limit: number = 10, stateFilter?: string): Promise<DbTaskInfo[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        const nodesSql = `SELECT id, name, updated_at FROM nodes WHERE type = 'TaskNode' ORDER BY updated_at DESC LIMIT ?`;
        const nodesRs: relationalStore.ResultSet = await this.store.querySql(nodesSql, [String(limit)]);
        const taskNodes: TaskNodeRow[] = [];
@ -827,27 +857,23 @@ export class GraphDatabase {
            while (relRs.goToNextRow()) {
                const rt: string = relRs.getString(relRs.getColumnIndex('relation'));
                const targetName: string = relRs.getString(relRs.getColumnIndex('target_name'));
-                if (rt === 'description')
-                    description = targetName;
-                if (rt === 'has_state')
-                    state = targetName;
+                if (rt === 'description') description = targetName;
+                if (rt === 'has_state') state = targetName;
            }
            relRs.close();
-            if (stateFilter && state !== stateFilter)
-                continue;
+            if (stateFilter && state !== stateFilter) continue;
            const cntSql = `SELECT COUNT(*) AS cnt FROM relations WHERE subject_id = ? AND relation = 'CONTAINS_INFO' AND status = 'active'`;
            const cntRs: relationalStore.ResultSet = await this.store.querySql(cntSql, [String(node.id)]);
            let infoCount = 0;
-            if (cntRs.goToNextRow())
-                infoCount = cntRs.getLong(cntRs.getColumnIndex('cnt'));
+            if (cntRs.goToNextRow()) infoCount = cntRs.getLong(cntRs.getColumnIndex('cnt'));
            cntRs.close();
            tasks.push({ taskId: node.name, description, state, infoCount, updatedAt: node.updatedAt });
        }
        return tasks;
    }
+
    private async removeOrphanNodes(): Promise<number> {
-        if (!this.store)
-            return 0;
+        if (!this.store) return 0;
        let deleted = 0;
        // 优化：批量查询所有有关系的节点 ID，避免 O(N²) 逐节点检查
        const activeRelPred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
@ -859,6 +885,7 @@ export class GraphDatabase {
            relatedIds.add(relResultSet.getLong(relResultSet.getColumnIndex('object_id')));
        }
        relResultSet.close();
+
        // 查询所有节点，筛选出不在关系中的孤儿节点
        const nodePred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        const nodeResultSet: relationalStore.ResultSet = await this.store.query(nodePred, ['id']);
@ -870,6 +897,7 @@ export class GraphDatabase {
            }
        }
        nodeResultSet.close();
+
        // 批量删除孤儿节点
        for (const orphanId of orphanIds) {
            const deletePred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
@ -879,9 +907,9 @@ export class GraphDatabase {
        }
        return deleted;
    }
+
    private async getNodeIdByName(name: string): Promise<number> {
-        if (!this.store)
-            return -1;
+        if (!this.store) return -1;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        predicates.equalTo('name', name);
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id']);
@ -893,9 +921,9 @@ export class GraphDatabase {
        resultSet.close();
        return -1;
    }
+
    async introspect(): Promise<IntrospectResult> {
-        if (!this.store)
-            return { entity_count: 0, relation_count: 0, message: 'Database not initialized' };
+        if (!this.store) return { entity_count: 0, relation_count: 0, message: 'Database not initialized' };
        const nodePredicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        const nodeResultSet = await this.store.query(nodePredicates, ['id']);
        let entityCount = 0;
@ -903,6 +931,7 @@ export class GraphDatabase {
            entityCount++;
        }
        nodeResultSet.close();
+
        const relPredicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        relPredicates.equalTo('status', 'active');
        const relResultSet = await this.store.query(relPredicates, ['id']);
@ -911,18 +940,20 @@ export class GraphDatabase {
            relationCount++;
        }
        relResultSet.close();
+
        return {
            entity_count: entityCount,
            relation_count: relationCount,
            message: `数据库包含 ${entityCount} 个实体, ${relationCount} 条关系`
        };
    }
+
    async archive(days: number): Promise<ArchiveResult> {
-        if (!this.store)
-            return { archived: 0, message: 'Database not initialized' };
+        if (!this.store) return { archived: 0, message: 'Database not initialized' };
        const cutoffDate = new Date();
        cutoffDate.setDate(cutoffDate.getDate() - days);
        const cutoffStr = cutoffDate.toISOString();
+
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        predicates.equalTo('status', 'active').and().lessThan('created_at', cutoffStr);
        const bucket: relationalStore.ValuesBucket = {
@ -930,19 +961,22 @@ export class GraphDatabase {
            'updated_at': new Date().toISOString()
        };
        const count = await this.store.update(bucket, predicates);
+
        return {
            archived: count,
            message: `归档了 ${count} 条关系`
        };
    }
+
    async cleanup(dryRun: boolean): Promise<CleanupResult> {
-        if (!this.store)
-            return { cleaned: 0, message: 'Database not initialized' };
+        if (!this.store) return { cleaned: 0, message: 'Database not initialized' };
        const cutoffDate = new Date();
        cutoffDate.setDate(cutoffDate.getDate() - 90);
        const cutoffStr = cutoffDate.toISOString();
+
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        predicates.equalTo('status', 'deleted').and().lessThan('updated_at', cutoffStr);
+
        let deleted = 0;
        if (dryRun) {
            const resultSet = await this.store.query(predicates, ['id']);
@ -950,8 +984,7 @@ export class GraphDatabase {
                deleted++;
            }
            resultSet.close();
-        }
-        else {
+        } else {
            deleted = await this.store.delete(predicates);
            const orphanCount = await this.removeOrphanNodes();
            return {
@ -962,6 +995,7 @@ export class GraphDatabase {
                message: `删除了 ${deleted} 条关系, ${orphanCount} 个孤立实体`
            } as CleanupResult;
        }
+
        return {
            cleaned: deleted,
            deleted_relations: deleted,
@ -969,9 +1003,9 @@ export class GraphDatabase {
            message: `将删除 ${deleted} 条关系`
        } as CleanupResult;
    }
+
    async saveChatMessage(role: string, content: string, tools?: string, sessionId?: string): Promise<void> {
-        if (!this.store)
-            return;
+        if (!this.store) return;
        const bucket: relationalStore.ValuesBucket = {
            'session_id': sessionId || null,
            'role': role,
@ -981,9 +1015,9 @@ export class GraphDatabase {
        };
        await this.store.insert('chat_records', bucket);
    }
+
    async getChatHistory(limit?: number, sessionId?: string): Promise<ChatMessage[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('chat_records');
        if (sessionId) {
            predicates.equalTo('session_id', sessionId);
@ -1005,15 +1039,15 @@ export class GraphDatabase {
        resultSet.close();
        return messages.reverse();
    }
+
    async clearChatHistory(): Promise<void> {
-        if (!this.store)
-            return;
+        if (!this.store) return;
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('chat_records');
        await this.store.delete(predicates);
    }
+
    private async getAllNodes(): Promise<NodeData[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
        const nodes: NodeData[] = [];
@ -1029,9 +1063,9 @@ export class GraphDatabase {
        resultSet.close();
        return nodes;
    }
+
    private async getAllEdges(): Promise<EdgeData[]> {
-        if (!this.store)
-            return [];
+        if (!this.store) return [];
        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'relation', 'object_id', 'weight']);
        const edges: EdgeData[] = [];
@ -1047,4 +1081,4 @@ export class GraphDatabase {
        resultSet.close();
        return edges;
    }
-}
+}
--- a/common/src/main/ets/model/GraphDatabase.ets.bak
+++ b/common/src/main/ets/model/GraphDatabase.ets.bak
@ -0,0 +1,936 @@
+import relationalStore from '@ohos.data.relationalStore';
+import { Context } from '@ohos.abilityAccessCtrl';
+
+interface NodeNameCacheItem { name: string; type: string; mentions: number; }
+export interface TimeRangeParams { days: number; }
+
+interface TripletData {
+    subject: string;
+    relation: string;
+    object: string;
+}
+
+interface CriteriaData {
+    subject?: string;
+    target?: string;
+    relation?: string;
+    sessionId?: string;
+}
+
+interface NodeData {
+    id: number;
+    label: string;
+    type: string;
+    mentions: number;
+    depth?: number;
+}
+
+interface EdgeData {
+    from: number;
+    to: number;
+    label: string;
+    weight: number;
+    depth?: number;
+    sessionId?: string;
+    turnId?: number;
+}
+
+interface GraphData {
+    nodes: NodeData[];
+    edges: EdgeData[];
+}
+
+export interface RecallEntity {
+    name: string;
+    type: string;
+    mention_count: number;
+    depth?: number;
+}
+
+interface RecallRelation {
+    source: string;
+    target: string;
+    type: string;
+    confidence: number;
+    session_id?: string;
+    turn_id?: number;
+    depth?: number;
+}
+
+interface RecallResult {
+    entities: RecallEntity[];
+    relations: RecallRelation[];
+    message: string;
+}
+
+interface BfsEntity {
+    id: number;
+    name: string;
+    type: string;
+    mentions: number;
+    depth: number;
+}
+
+export interface RelationQueryResult {
+    sourceId: number;
+    targetId: number;
+    sourceName: string;
+    targetName: string;
+    type: string;
+    confidence: number;
+    sessionId?: string;
+    turnId?: number;
+    depth: number;
+}
+
+interface NodeQueryResult {
+    id: number;
+    name: string;
+    type: string;
+    mentions: number;
+    depth: number;
+}
+
+interface CleanupResult {
+    cleaned: number;
+    deleted_relations?: number;
+    deleted_orphans?: number;
+    dry_run?: boolean;
+    message?: string;
+}
+
+interface IntrospectResult {
+    entity_count: number;
+    relation_count: number;
+    message: string;
+}
+
+interface ArchiveResult {
+    archived: number;
+    message: string;
+}
+
+interface SearchResultItem {
+    name: string;
+    type: string;
+    mentions: number;
+}
+
+interface ChatMessage {
+    role: string;
+    content: string;
+    session_id?: string;
+}
+
+interface SnapshotData {
+    entities: RecallEntity[];
+    relations: RecallRelation[];
+}
+
+const STORE_CONFIG: relationalStore.StoreConfig = {
+    name: 'trulymem.db',
+    securityLevel: relationalStore.SecurityLevel.S1
+};
+
+export class GraphDatabase {
+    private store?: relationalStore.RdbStore;
+    private context?: Context;
+
+    async init(context: Context): Promise<void> {
+        this.context = context;
+        this.store = await relationalStore.getRdbStore(context, STORE_CONFIG);
+        await this.createTables();
+    }
+
+    private async createTables(): Promise<void> {
+        if (!this.store) return;
+        await this.store.executeSql(`
+            CREATE TABLE IF NOT EXISTS nodes (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                name TEXT NOT NULL UNIQUE,
+                type TEXT DEFAULT 'concept',
+                mentions INTEGER DEFAULT 1,
+                created_at TEXT,
+                updated_at TEXT
+            )
+        `);
+        await this.store.executeSql(`
+            CREATE TABLE IF NOT EXISTS relations (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                subject_id INTEGER NOT NULL,
+                relation TEXT NOT NULL,
+                object_id INTEGER NOT NULL,
+                weight REAL DEFAULT 1.0,
+                session_id TEXT,
+                turn_id INTEGER,
+                created_at TEXT,
+                updated_at TEXT,
+                status TEXT DEFAULT 'active',
+                date_bucket TEXT
+            )
+        `);
+        await this.store.executeSql(`
+            CREATE TABLE IF NOT EXISTS chat_records (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                session_id TEXT,
+                role TEXT NOT NULL,
+                content TEXT NOT NULL,
+                tools TEXT,
+                created_at TEXT
+            )
+        `);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_node_name ON nodes(name)`);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_node_type ON nodes(type)`);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_source ON relations(subject_id)`);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_target ON relations(object_id)`);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_type ON relations(relation)`);
+        await this.store.executeSql(`CREATE INDEX IF NOT EXISTS idx_rel_status ON relations(status)`);
+    }
+
+    async commit(triplets: TripletData[], entityTypes?: Record<string, string>, sessionId?: string, turnId?: number): Promise<void> {
+        if (!this.store) return;
+        for (const triplet of triplets) {
+            const subjectId: number = await this.upsertNode(triplet.subject, entityTypes?.[triplet.subject]);
+            const objectId: number = await this.upsertNode(triplet.object, entityTypes?.[triplet.object]);
+            const existingId: number = await this.checkDuplicateRelation(subjectId, triplet.relation, objectId);
+            if (existingId > 0) {
+                continue;
+            }
+            const now = new Date().toISOString();
+            const dateBucket = now.split('T')[0];
+            const bucket: relationalStore.ValuesBucket = {
+                'subject_id': subjectId,
+                'relation': triplet.relation,
+                'object_id': objectId,
+                'session_id': sessionId || null,
+                'turn_id': turnId || null,
+                'created_at': now,
+                'updated_at': now,
+                'status': 'active',
+                'date_bucket': dateBucket
+            };
+            await this.store.insert('relations', bucket);
+        }
+    }
+
+    private async checkDuplicateRelation(subjectId: number, relation: string, objectId: number): Promise<number> {
+        if (!this.store) return -1;
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('subject_id', subjectId).and().equalTo('relation', relation).and().equalTo('object_id', objectId).and().equalTo('status', 'active');
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id']);
+        if (resultSet.goToFirstRow()) {
+            const id: number = resultSet.getLong(resultSet.getColumnIndex('id'));
+            resultSet.close();
+            return id;
+        }
+        resultSet.close();
+        return -1;
+    }
+
+    private async upsertNode(name: string, entityType?: string): Promise<number> {
+        if (!this.store) return -1;
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        predicates.equalTo('name', name);
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'mentions']);
+        const now = new Date().toISOString();
+        if (resultSet.goToNextRow()) {
+            const id: number = resultSet.getLong(resultSet.getColumnIndex('id'));
+            const mentions: number = resultSet.getLong(resultSet.getColumnIndex('mentions'));
+            resultSet.close();
+            const updatePredicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+            updatePredicates.equalTo('id', id);
+            const bucket: relationalStore.ValuesBucket = {
+                'mentions': mentions + 1,
+                'updated_at': now
+            };
+            await this.store.update(bucket, updatePredicates);
+            return id;
+        }
+        resultSet.close();
+        const bucket: relationalStore.ValuesBucket = {
+            'name': name,
+            'type': entityType || 'concept',
+            'mentions': 1,
+            'created_at': now,
+            'updated_at': now
+        };
+        return await this.store.insert('nodes', bucket);
+    }
+
+    async recall(queryIntent: string, seedEntities?: string[], depth: number = 2, timeRange?: TimeRangeParams, sessionFilter?: string): Promise<RecallResult> {
+        if (!this.store) {
+            return { entities: [], relations: [], message: 'Database not initialized' };
+        }
+        // 计算时间范围过滤
+        let minDateBucket: string | undefined;
+        if (timeRange && timeRange.days && timeRange.days > 0) {
+            const cutoff = new Date();
+            cutoff.setDate(cutoff.getDate() - timeRange.days);
+            minDateBucket = cutoff.toISOString().slice(0, 10).replace(/-/g, '');
+        }
+        const keywords = queryIntent.toLowerCase().replace(/,/g, ' ').split(/\s+/).filter(w => w.trim());
+        const allEntities: BfsEntity[] = [];
+        const entityIds = new Set<number>();
+        let seedEntityIds = new Set<number>();
+
+        if (keywords.length === 0 && (!seedEntities || seedEntities.length === 0)) {
+            const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+            predicates.orderByDesc('mentions');
+            const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
+            while (resultSet.goToNextRow() && allEntities.length < 50) {
+                const id = resultSet.getLong(resultSet.getColumnIndex('id'));
+                const name = resultSet.getString(resultSet.getColumnIndex('name'));
+                const type = resultSet.getString(resultSet.getColumnIndex('type'));
+                const mentions = resultSet.getLong(resultSet.getColumnIndex('mentions'));
+                entityIds.add(id);
+                allEntities.push({ id, name, type, mentions, depth: 0 });
+            }
+            resultSet.close();
+        } else {
+            if (seedEntities && seedEntities.length > 0) {
+                for (const seedName of seedEntities) {
+                    const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+                    predicates.equalTo('name', seedName);
+                    const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
+                    while (resultSet.goToNextRow()) {
+                        const id = resultSet.getLong(resultSet.getColumnIndex('id'));
+                        if (!entityIds.has(id)) {
+                            entityIds.add(id);
+                            seedEntityIds.add(id);
+                            allEntities.push({
+                                id,
+                                name: resultSet.getString(resultSet.getColumnIndex('name')),
+                                type: resultSet.getString(resultSet.getColumnIndex('type')),
+                                mentions: resultSet.getLong(resultSet.getColumnIndex('mentions')),
+                                depth: 0
+                            });
+                        }
+                    }
+                    resultSet.close();
+                }
+            }
+            for (const keyword of keywords) {
+                const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+                predicates.like('name', `%${keyword}%`);
+                const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
+                while (resultSet.goToNextRow()) {
+                    const id = resultSet.getLong(resultSet.getColumnIndex('id'));
+                    if (!entityIds.has(id)) {
+                        entityIds.add(id);
+                        allEntities.push({
+                            id,
+                            name: resultSet.getString(resultSet.getColumnIndex('name')),
+                            type: resultSet.getString(resultSet.getColumnIndex('type')),
+                            mentions: resultSet.getLong(resultSet.getColumnIndex('mentions')),
+                            depth: 0
+                        });
+                    }
+                }
+                resultSet.close();
+            }
+        }
+
+        const allRelations: RelationQueryResult[] = [];
+        let currentLayerIds = new Set<number>(entityIds);
+        const visitedEntityIds = new Set<number>(entityIds);
+        // 批量预加载所有相关节点名称，减少 N+1 查询
+        const nodeNameCache = new Map<number, NodeNameCacheItem>();
+
+        for (let layer = 0; layer < depth && currentLayerIds.size > 0; layer++) {
+            const currentIds = Array.from(currentLayerIds);
+            const relations = await this.getRelationsForNodes(currentIds, sessionFilter, minDateBucket);
+            const nextLayerIds = new Set<number>();
+
+            for (const rel of relations) {
+                allRelations.push(rel);
+                if (!visitedEntityIds.has(rel.targetId)) {
+                    nextLayerIds.add(rel.targetId);
+                }
+                if (rel.targetId !== rel.sourceId && !visitedEntityIds.has(rel.sourceId)) {
+                    nextLayerIds.add(rel.sourceId);
+                }
+            }
+
+            for (const newId of nextLayerIds) {
+                if (!visitedEntityIds.has(newId)) {
+                    visitedEntityIds.add(newId);
+                    // 优先从缓存获取，避免 N+1 查询
+                    const cached = nodeNameCache.get(newId);
+                    if (cached) {
+                        const addedEntity: BfsEntity = {
+                            id: newId,
+                            name: cached.name,
+                            type: cached.type,
+                            mentions: cached.mentions,
+                            depth: layer + 1
+                        };
+                        allEntities.push(addedEntity);
+                    } else {
+                        const nodeData = await this.getNodeById(newId);
+                        if (nodeData) {
+                            const cacheItem: NodeNameCacheItem = { name: nodeData.name, type: nodeData.type, mentions: nodeData.mentions };
+                            nodeNameCache.set(newId, cacheItem);
+                            const addedEntity: BfsEntity = {
+                                id: nodeData.id,
+                                name: nodeData.name,
+                                type: nodeData.type,
+                                mentions: nodeData.mentions,
+                                depth: layer + 1
+                            };
+                            allEntities.push(addedEntity);
+                        }
+                    }
+                }
+            }
+            currentLayerIds = nextLayerIds;
+        }
+
+        const entities: RecallEntity[] = allEntities.map(e => {
+            const entity: RecallEntity = {
+                name: e.name,
+                type: e.type,
+                mention_count: e.mentions,
+                depth: e.depth
+            };
+            return entity;
+        });
+        const relations: RecallRelation[] = allRelations.map(r => {
+            const rel: RecallRelation = {
+                source: r.sourceName,
+                target: r.targetName,
+                type: r.type,
+                confidence: r.confidence,
+                session_id: r.sessionId,
+                turn_id: r.turnId,
+                depth: r.depth
+            };
+            return rel;
+        });
+
+        return {
+            entities,
+            relations,
+            message: `找到 ${entities.length} 个实体, ${relations.length} 条关系`
+        };
+    }
+
+    private async getNodeById(id: number): Promise<NodeQueryResult | null> {
+        if (!this.store) return null;
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        predicates.equalTo('id', id);
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
+        if (resultSet.goToNextRow()) {
+            const node: NodeQueryResult = {
+                id: resultSet.getLong(resultSet.getColumnIndex('id')),
+                name: resultSet.getString(resultSet.getColumnIndex('name')),
+                type: resultSet.getString(resultSet.getColumnIndex('type')),
+                mentions: resultSet.getLong(resultSet.getColumnIndex('mentions')),
+                depth: 0
+            };
+            resultSet.close();
+            return node;
+        }
+        resultSet.close();
+        return null;
+    }
+
+    private async getRelationsForNodes(nodeIds: number[], sessionFilter?: string, minDateBucket?: string): Promise<RelationQueryResult[]> {
+        if (!this.store || nodeIds.length === 0) return [];
+        const relations: RelationQueryResult[] = [];
+        // 批量预加载所有节点名称到缓存，避免 N+1 查询
+        const nodeNameCache = new Map<number, NodeNameCacheItem>();
+        for (const id of nodeIds) {
+            const node = await this.getNodeById(id);
+            if (node) {
+                const cacheItem: NodeNameCacheItem = { name: node.name, type: node.type, mentions: node.mentions };
+                nodeNameCache.set(id, cacheItem);
+            }
+        }
+        for (const nodeId of nodeIds) {
+            const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+            predicates.equalTo('status', 'active').and().equalTo('subject_id', nodeId);
+            if (sessionFilter) {
+                predicates.and().equalTo('session_id', sessionFilter);
+            }
+            if (minDateBucket) {
+                predicates.and().greaterThanOrEqualTo('date_bucket', minDateBucket);
+            }
+            const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'object_id', 'relation', 'weight', 'session_id', 'turn_id']);
+            while (resultSet.goToNextRow()) {
+                const sourceId = resultSet.getLong(resultSet.getColumnIndex('subject_id'));
+                const targetId = resultSet.getLong(resultSet.getColumnIndex('object_id'));
+                const sourceNode = nodeNameCache.get(sourceId) || await this.getNodeById(sourceId);
+                const targetNode = nodeNameCache.get(targetId) || await this.getNodeById(targetId);
+                if (sourceNode && targetNode) {
+                    relations.push({
+                        sourceId,
+                        targetId,
+                        sourceName: sourceNode.name,
+                        targetName: targetNode.name,
+                        type: resultSet.getString(resultSet.getColumnIndex('relation')),
+                        confidence: resultSet.getDouble(resultSet.getColumnIndex('weight')),
+                        sessionId: resultSet.getString(resultSet.getColumnIndex('session_id')),
+                        turnId: resultSet.getLong(resultSet.getColumnIndex('turn_id')),
+                        depth: 1
+                    });
+                }
+            }
+            resultSet.close();
+
+            const predicates2: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+            predicates2.equalTo('status', 'active').and().equalTo('object_id', nodeId);
+            if (sessionFilter) {
+                predicates2.and().equalTo('session_id', sessionFilter);
+            }
+            if (minDateBucket) {
+                predicates2.and().greaterThanOrEqualTo('date_bucket', minDateBucket);
+            }
+            const resultSet2: relationalStore.ResultSet = await this.store.query(predicates2, ['subject_id', 'object_id', 'relation', 'weight', 'session_id', 'turn_id']);
+            while (resultSet2.goToNextRow()) {
+                const sourceId = resultSet2.getLong(resultSet2.getColumnIndex('subject_id'));
+                const targetId = resultSet2.getLong(resultSet2.getColumnIndex('object_id'));
+                const sourceNode = nodeNameCache.get(sourceId) || await this.getNodeById(sourceId);
+                const targetNode = nodeNameCache.get(targetId) || await this.getNodeById(targetId);
+                if (sourceNode && targetNode) {
+                    relations.push({
+                        sourceId,
+                        targetId,
+                        sourceName: sourceNode.name,
+                        targetName: targetNode.name,
+                        type: resultSet2.getString(resultSet2.getColumnIndex('relation')),
+                        confidence: resultSet2.getDouble(resultSet2.getColumnIndex('weight')),
+                        sessionId: resultSet2.getString(resultSet2.getColumnIndex('session_id')),
+                        turnId: resultSet2.getLong(resultSet2.getColumnIndex('turn_id')),
+                        depth: 1
+                    });
+                }
+            }
+            resultSet2.close();
+        }
+        return relations;
+    }
+
+    async search(keyword: string): Promise<SearchResultItem[]> {
+        if (!this.store) return [];
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        predicates.like('name', `%${keyword}%`);
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['name', 'type', 'mentions']);
+        const results: SearchResultItem[] = [];
+        while (resultSet.goToNextRow()) {
+            results.push({
+                name: resultSet.getString(resultSet.getColumnIndex('name')),
+                type: resultSet.getString(resultSet.getColumnIndex('type')),
+                mentions: resultSet.getLong(resultSet.getColumnIndex('mentions'))
+            });
+        }
+        resultSet.close();
+        return results;
+    }
+
+    async purge(criteria: CriteriaData, mode: string = 'soft'): Promise<void> {
+        if (!this.store) return;
+        if (!criteria.subject && !criteria.target && !criteria.relation && !criteria.sessionId) {
+            return;
+        }
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        let hasCondition = false;
+
+        if (criteria.subject) {
+            const subjectId = await this.getNodeIdByName(criteria.subject);
+            if (subjectId > 0) {
+                predicates.equalTo('subject_id', subjectId);
+                hasCondition = true;
+            }
+        }
+        if (criteria.target) {
+            const targetId = await this.getNodeIdByName(criteria.target);
+            if (targetId > 0) {
+                if (hasCondition) {
+                    predicates.and();
+                }
+                predicates.equalTo('object_id', targetId);
+                hasCondition = true;
+            }
+        }
+        if (criteria.relation) {
+            if (hasCondition) {
+                predicates.and();
+            }
+            predicates.equalTo('relation', criteria.relation);
+            hasCondition = true;
+        }
+        if (criteria.sessionId) {
+            if (hasCondition) {
+                predicates.and();
+            }
+            predicates.equalTo('session_id', criteria.sessionId);
+            hasCondition = true;
+        }
+
+        if (hasCondition) {
+            if (mode === 'soft') {
+                const bucket: relationalStore.ValuesBucket = {
+                    'status': 'deleted',
+                    'updated_at': new Date().toISOString()
+                };
+                await this.store.update(bucket, predicates);
+            } else {
+                await this.store.delete(predicates);
+            }
+        }
+        await this.removeOrphanNodes();
+    }
+
+    /**
+     * 记忆图谱 — 在指定时间范围内查询关系和节点
+     * 对应 tools.memory_graph
+     */
+    async graph(timeRange: TimeRangeParams, sessionFilter?: string): Promise<GraphData> {
+        if (!this.store) return { nodes: [], edges: [] };
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('status', 'active');
+        if (sessionFilter) {
+            predicates.and().equalTo('session_id', sessionFilter);
+        }
+        if (timeRange && timeRange.days && timeRange.days > 0) {
+            const cutoff = new Date();
+            cutoff.setDate(cutoff.getDate() - timeRange.days);
+            const minDateBucket = cutoff.toISOString().slice(0, 10).replace(/-/g, '');
+            predicates.and().greaterThanOrEqualTo('date_bucket', minDateBucket);
+        }
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'object_id', 'relation', 'weight', 'session_id', 'turn_id']);
+        const nodeIds = new Set<number>();
+        const edges: EdgeData[] = [];
+        while (resultSet.goToNextRow()) {
+            const sourceId = resultSet.getLong(resultSet.getColumnIndex('subject_id'));
+            const targetId = resultSet.getLong(resultSet.getColumnIndex('object_id'));
+            nodeIds.add(sourceId);
+            nodeIds.add(targetId);
+            const edge: EdgeData = {
+                from: sourceId,
+                to: targetId,
+                label: resultSet.getString(resultSet.getColumnIndex('relation')),
+                weight: resultSet.getDouble(resultSet.getColumnIndex('weight')),
+                sessionId: resultSet.getString(resultSet.getColumnIndex('session_id')),
+                turnId: resultSet.getLong(resultSet.getColumnIndex('turn_id'))
+            };
+            edges.push(edge);
+        }
+        resultSet.close();
+        const nodes: NodeData[] = [];
+        for (const id of nodeIds) {
+            const node = await this.getNodeById(id);
+            if (node) {
+                const nodeData: NodeData = { id: node.id, label: node.name, type: node.type, mentions: node.mentions };
+                nodes.push(nodeData);
+            }
+        }
+        const result: GraphData = { nodes, edges };
+        return result;
+    }
+
+    /**
+     * 记忆快照 — 在指定时间范围内查询实体和关系
+     * 对应 tools.memory_snapshot
+     */
+    async snapshot(timeRange: TimeRangeParams, sessionFilter?: string): Promise<SnapshotData> {
+        if (!this.store) return { entities: [], relations: [] };
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('status', 'active');
+        if (sessionFilter) {
+            predicates.and().equalTo('session_id', sessionFilter);
+        }
+        if (timeRange && timeRange.days && timeRange.days > 0) {
+            const cutoff = new Date();
+            cutoff.setDate(cutoff.getDate() - timeRange.days);
+            const minDateBucket = cutoff.toISOString().slice(0, 10).replace(/-/g, '');
+            predicates.and().greaterThanOrEqualTo('date_bucket', minDateBucket);
+        }
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'object_id', 'relation', 'weight', 'session_id', 'turn_id']);
+        const nodeIds = new Set<number>();
+        const relations: RecallRelation[] = [];
+        while (resultSet.goToNextRow()) {
+            const sourceId = resultSet.getLong(resultSet.getColumnIndex('subject_id'));
+            const targetId = resultSet.getLong(resultSet.getColumnIndex('object_id'));
+            nodeIds.add(sourceId);
+            nodeIds.add(targetId);
+            const sourceNode = await this.getNodeById(sourceId);
+            const targetNode = await this.getNodeById(targetId);
+            if (sourceNode && targetNode) {
+                const rel: RecallRelation = {
+                    source: sourceNode.name,
+                    target: targetNode.name,
+                    type: resultSet.getString(resultSet.getColumnIndex('relation')),
+                    confidence: resultSet.getDouble(resultSet.getColumnIndex('weight')),
+                    session_id: resultSet.getString(resultSet.getColumnIndex('session_id')),
+                    turn_id: resultSet.getLong(resultSet.getColumnIndex('turn_id'))
+                };
+                relations.push(rel);
+            }
+        }
+        resultSet.close();
+        const entities: RecallEntity[] = [];
+        for (const id of nodeIds) {
+            const node = await this.getNodeById(id);
+            if (node) {
+                const recallEntity: RecallEntity = { name: node.name, type: node.type, mention_count: node.mentions };
+                entities.push(recallEntity);
+            }
+        }
+        const snapResult: SnapshotData = { entities, relations };
+        return snapResult;
+    }
+
+    /**
+     * 查询已归档的记忆
+     * 对应 tools.memory_query_archived
+     */
+    async queryArchived(days?: number, keyword?: string): Promise<RelationQueryResult[]> {
+        if (!this.store) return [];
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('status', 'archived');
+        if (keyword) {
+            predicates.and().like('relation', '%' + keyword + '%');
+        }
+        if (days && days > 0) {
+            const cutoff = new Date();
+            cutoff.setDate(cutoff.getDate() - days);
+            const maxDateBucket = cutoff.toISOString().slice(0, 10).replace(/-/g, '');
+            predicates.and().lessThanOrEqualTo('date_bucket', maxDateBucket);
+        }
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'object_id', 'relation', 'weight', 'session_id', 'turn_id']);
+        const results: RelationQueryResult[] = [];
+        while (resultSet.goToNextRow()) {
+            const sourceId = resultSet.getLong(resultSet.getColumnIndex('subject_id'));
+            const targetId = resultSet.getLong(resultSet.getColumnIndex('object_id'));
+            const sourceNode = await this.getNodeById(sourceId);
+            const targetNode = await this.getNodeById(targetId);
+            if (sourceNode && targetNode) {
+                const queryResult: RelationQueryResult = {
+                    sourceId: sourceId,
+                    targetId: targetId,
+                    sourceName: sourceNode.name,
+                    targetName: targetNode.name,
+                    type: resultSet.getString(resultSet.getColumnIndex('relation')),
+                    confidence: resultSet.getDouble(resultSet.getColumnIndex('weight')),
+                    sessionId: resultSet.getString(resultSet.getColumnIndex('session_id')),
+                    turnId: resultSet.getLong(resultSet.getColumnIndex('turn_id')),
+                    depth: 0
+                };
+                results.push(queryResult);
+            }
+        }
+        resultSet.close();
+        return results;
+    }
+
+    private async removeOrphanNodes(): Promise<number> {
+        if (!this.store) return 0;
+        let deleted = 0;
+        // 优化：批量查询所有有关系的节点 ID，避免 O(N²) 逐节点检查
+        const activeRelPred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        activeRelPred.equalTo('status', 'active');
+        const relResultSet: relationalStore.ResultSet = await this.store.query(activeRelPred, ['subject_id', 'object_id']);
+        const relatedIds = new Set<number>();
+        while (relResultSet.goToNextRow()) {
+            relatedIds.add(relResultSet.getLong(relResultSet.getColumnIndex('subject_id')));
+            relatedIds.add(relResultSet.getLong(relResultSet.getColumnIndex('object_id')));
+        }
+        relResultSet.close();
+
+        // 查询所有节点，筛选出不在关系中的孤儿节点
+        const nodePred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        const nodeResultSet: relationalStore.ResultSet = await this.store.query(nodePred, ['id']);
+        const orphanIds: number[] = [];
+        while (nodeResultSet.goToNextRow()) {
+            const nodeId = nodeResultSet.getLong(nodeResultSet.getColumnIndex('id'));
+            if (!relatedIds.has(nodeId)) {
+                orphanIds.push(nodeId);
+            }
+        }
+        nodeResultSet.close();
+
+        // 批量删除孤儿节点
+        for (const orphanId of orphanIds) {
+            const deletePred: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+            deletePred.equalTo('id', orphanId);
+            await this.store.delete(deletePred);
+            deleted++;
+        }
+        return deleted;
+    }
+
+    private async getNodeIdByName(name: string): Promise<number> {
+        if (!this.store) return -1;
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        predicates.equalTo('name', name);
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id']);
+        if (resultSet.goToNextRow()) {
+            const id = resultSet.getLong(resultSet.getColumnIndex('id'));
+            resultSet.close();
+            return id;
+        }
+        resultSet.close();
+        return -1;
+    }
+
+    async introspect(): Promise<IntrospectResult> {
+        if (!this.store) return { entity_count: 0, relation_count: 0, message: 'Database not initialized' };
+        const nodePredicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        const nodeResultSet = await this.store.query(nodePredicates, ['id']);
+        let entityCount = 0;
+        while (nodeResultSet.goToNextRow()) {
+            entityCount++;
+        }
+        nodeResultSet.close();
+
+        const relPredicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        relPredicates.equalTo('status', 'active');
+        const relResultSet = await this.store.query(relPredicates, ['id']);
+        let relationCount = 0;
+        while (relResultSet.goToNextRow()) {
+            relationCount++;
+        }
+        relResultSet.close();
+
+        return {
+            entity_count: entityCount,
+            relation_count: relationCount,
+            message: `数据库包含 ${entityCount} 个实体, ${relationCount} 条关系`
+        };
+    }
+
+    async archive(days: number): Promise<ArchiveResult> {
+        if (!this.store) return { archived: 0, message: 'Database not initialized' };
+        const cutoffDate = new Date();
+        cutoffDate.setDate(cutoffDate.getDate() - days);
+        const cutoffStr = cutoffDate.toISOString();
+
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('status', 'active').and().lessThan('created_at', cutoffStr);
+        const bucket: relationalStore.ValuesBucket = {
+            'status': 'archived',
+            'updated_at': new Date().toISOString()
+        };
+        const count = await this.store.update(bucket, predicates);
+
+        return {
+            archived: count,
+            message: `归档了 ${count} 条关系`
+        };
+    }
+
+    async cleanup(dryRun: boolean): Promise<CleanupResult> {
+        if (!this.store) return { cleaned: 0, message: 'Database not initialized' };
+        const cutoffDate = new Date();
+        cutoffDate.setDate(cutoffDate.getDate() - 90);
+        const cutoffStr = cutoffDate.toISOString();
+
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        predicates.equalTo('status', 'deleted').and().lessThan('updated_at', cutoffStr);
+
+        let deleted = 0;
+        if (dryRun) {
+            const resultSet = await this.store.query(predicates, ['id']);
+            while (resultSet.goToNextRow()) {
+                deleted++;
+            }
+            resultSet.close();
+        } else {
+            deleted = await this.store.delete(predicates);
+            const orphanCount = await this.removeOrphanNodes();
+            return {
+                cleaned: deleted + orphanCount,
+                deleted_relations: deleted,
+                deleted_orphans: orphanCount,
+                dry_run: false,
+                message: `删除了 ${deleted} 条关系, ${orphanCount} 个孤立实体`
+            } as CleanupResult;
+        }
+
+        return {
+            cleaned: deleted,
+            deleted_relations: deleted,
+            dry_run: true,
+            message: `将删除 ${deleted} 条关系`
+        } as CleanupResult;
+    }
+
+    async saveChatMessage(role: string, content: string, tools?: string, sessionId?: string): Promise<void> {
+        if (!this.store) return;
+        const bucket: relationalStore.ValuesBucket = {
+            'session_id': sessionId || null,
+            'role': role,
+            'content': content,
+            'tools': tools || null,
+            'created_at': new Date().toISOString()
+        };
+        await this.store.insert('chat_records', bucket);
+    }
+
+    async getChatHistory(limit?: number, sessionId?: string): Promise<ChatMessage[]> {
+        if (!this.store) return [];
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('chat_records');
+        if (sessionId) {
+            predicates.equalTo('session_id', sessionId);
+        }
+        predicates.orderByDesc('created_at');
+        if (limit) {
+            predicates.limitAs(limit);
+        }
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['role', 'content', 'session_id']);
+        const messages: ChatMessage[] = [];
+        while (resultSet.goToNextRow()) {
+            const msg: ChatMessage = {
+                role: resultSet.getString(resultSet.getColumnIndex('role')),
+                content: resultSet.getString(resultSet.getColumnIndex('content')),
+                session_id: resultSet.getString(resultSet.getColumnIndex('session_id'))
+            };
+            messages.push(msg);
+        }
+        resultSet.close();
+        return messages.reverse();
+    }
+
+    async clearChatHistory(): Promise<void> {
+        if (!this.store) return;
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('chat_records');
+        await this.store.delete(predicates);
+    }
+
+    private async getAllNodes(): Promise<NodeData[]> {
+        if (!this.store) return [];
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('nodes');
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['id', 'name', 'type', 'mentions']);
+        const nodes: NodeData[] = [];
+        while (resultSet.goToNextRow()) {
+            const node: NodeData = {
+                id: resultSet.getLong(resultSet.getColumnIndex('id')),
+                label: resultSet.getString(resultSet.getColumnIndex('name')),
+                type: resultSet.getString(resultSet.getColumnIndex('type')),
+                mentions: resultSet.getLong(resultSet.getColumnIndex('mentions'))
+            };
+            nodes.push(node);
+        }
+        resultSet.close();
+        return nodes;
+    }
+
+    private async getAllEdges(): Promise<EdgeData[]> {
+        if (!this.store) return [];
+        const predicates: relationalStore.RdbPredicates = new relationalStore.RdbPredicates('relations');
+        const resultSet: relationalStore.ResultSet = await this.store.query(predicates, ['subject_id', 'relation', 'object_id', 'weight']);
+        const edges: EdgeData[] = [];
+        while (resultSet.goToNextRow()) {
+            const edge: EdgeData = {
+                from: resultSet.getLong(resultSet.getColumnIndex('subject_id')),
+                to: resultSet.getLong(resultSet.getColumnIndex('object_id')),
+                label: resultSet.getString(resultSet.getColumnIndex('relation')),
+                weight: resultSet.getDouble(resultSet.getColumnIndex('weight'))
+            };
+            edges.push(edge);
+        }
+        resultSet.close();
+        return edges;
+    }
+}
--- a/common/src/main/ets/routermanager/PageContext.ets
+++ b/common/src/main/ets/routermanager/PageContext.ets
@ -0,0 +1,56 @@
+import { defaultLogger } from '../util/Logger';
+
+export interface RouterParam {
+  routerName: string;
+  param?: object;
+}
+
+export interface IPageContext {
+  openPage(data: RouterParam, animated?: boolean): void;
+  popPage(animated?: boolean): void;
+  replacePage(data: RouterParam, animated?: boolean): void;
+}
+
+export class PageContext implements IPageContext {
+  private readonly pathStack: NavPathStack;
+
+  constructor() {
+    this.pathStack = new NavPathStack();
+  }
+
+  public get navPathStack(): NavPathStack {
+    return this.pathStack;
+  }
+
+  public replacePage(data: RouterParam, animated: boolean = true): void {
+    try {
+      this.pathStack.replacePath({ name: data.routerName, param: data.param }, animated);
+    } catch (err) {
+      defaultLogger.error('replacePage: ' + data.routerName + ' failed. ' + err.code + ' ' + err.message);
+    }
+  }
+
+  public openPage(data: RouterParam, animated: boolean = true): void {
+    try {
+      this.pathStack.pushPath({ name: data.routerName, param: data.param }, animated);
+    } catch (err) {
+      defaultLogger.error('openPage: ' + data.routerName + ' failed. ' + err.code + ' ' + err.message);
+    }
+  }
+
+  public popPage(animated: boolean = true): void {
+    try {
+      this.pathStack.pop(animated);
+    } catch (err) {
+      defaultLogger.error('popPage failed. ' + err.code + ' ' + err.message);
+    }
+  }
+
+  public popPageByIndex(index: number, animated: boolean = true): void {
+    this.pathStack.popToIndex(index, animated);
+  }
+
+  public clear(animated: boolean = true): void {
+    this.pathStack.clear(animated);
+  }
+}
--- a/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/service/AIAgentService.ts
+++ b/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/service/AIAgentService.ts
@ -1,23 +1,33 @@
-import http from "@ohos:net.http";
-import dataPreferences from "@ohos:data.preferences";
-import type common from "@ohos:app.ability.common";
-import { defaultLogger } from "@normalized:N&&&@ohos/common/src/main/ets/util/Logger&1.0.0";
-import type { GraphMemoryService, EntityInfo, RelationInfo, MemoryRecallParams, MemoryCommitParams, MemoryPurgeParams, PurgeCriteriaParams, NewRelationParams, PersonaUpdateParams, TaskCreateParams, TaskSetStateParams, TaskDeleteParams, TaskLinkInfoParams, TaskArchiveParams, TripletInput, PersonaQueryResult, MemoryRecallResult } from './GraphMemoryService';
-import type { TimeRangeParams } from '../model/GraphDatabase';
+/**
+ * AIAgentService - AI Agent 服务层
+ * 管理上下文感知的 AI 对话，注入图数据作为上下文，
+ * 解析 AI 返回中的记忆操作，调用 GraphMemoryService 执行
+ * 参考：main 分支 core/graph_client.py
+ */
+import http from '@ohos.net.http';
+import dataPreferences from '@ohos.data.preferences';
+import { Context } from '@ohos.abilityAccessCtrl';
+import { GraphMemoryService, EntityInfo, RelationInfo, TaskInfo, MemoryRecallParams, MemoryCommitParams, MemoryPurgeParams, PurgeCriteriaParams, NewRelationParams, PersonaUpdateParams, TaskCreateParams, TaskSetStateParams, TaskDeleteParams, TaskLinkInfoParams, TaskArchiveParams, TaskQueryParams, TripletInput, PersonaQueryResult, TaskQueryResult, MemoryRecallResult } from './GraphMemoryService';
+import { TimeRangeParams } from '../model/GraphDatabase';
+
 export interface ChatMessage {
    role: string;
    content: string;
 }
+
 export interface AgentResponse {
    content: string;
    toolCalls: ToolCallResult[];
 }
+
 export interface ToolCallResult {
    name: string;
    success: boolean;
    message: string;
 }
+
 // ========= 工具定义类型 =========
+
 // Concrete interface for tool property definitions (replaces Record<string, T>)
 interface ToolPropertiesDefinition {
    days?: ToolParamProperty;
@ -62,6 +72,7 @@ interface ToolPropertiesDefinition {
    targetType?: ToolParamProperty;
    sourceHasStatus?: ToolParamProperty;
 }
+
 interface ToolParamProperty {
    type: string;
    description: string;
@ -70,56 +81,74 @@ interface ToolParamProperty {
    required?: string[];
    enum?: string[];
 }
+
 interface ToolParamDecl {
    type: string;
    properties: ToolPropertiesDefinition;
    required?: string[];
 }
+
 interface ToolFunctionDecl {
    name: string;
    description: string;
    parameters: ToolParamDecl;
 }
+
 interface ToolFunctionDef {
    type: string;
    function: ToolFunctionDecl;
 }
+
 // ========= API 请求/响应结构 =========
+
 interface ApiRequestMessage {
    role: string;
    content: string;
 }
+
 interface ApiRequest {
    model: string;
    messages: ApiRequestMessage[];
    tools?: ToolFunctionDef[];
    tool_choice?: string;
 }
+
 interface ApiToolCall {
    id: string;
    type: string;
    function: ToolFunctionCall;
 }
+
 interface ToolFunctionCall {
    name: string;
    arguments: string;
 }
+
 interface ApiChoiceMessage {
    content?: string;
    tool_calls?: ApiToolCall[];
 }
+
 interface ApiChoice {
    message: ApiChoiceMessage;
 }
+
 interface ApiResponse {
    choices: ApiChoice[];
 }
+
 // ========= 内部结果类型 =========
+
 interface ExecuteToolResult {
    name: string;
    success: boolean;
    message: string;
 }
+
+
+
+
+
 interface BuildContextBlockParams {
    persona: Record<string, string>;
    found: boolean;
@ -127,16 +156,23 @@ interface BuildContextBlockParams {
    relations: RelationInfo[];
    message: string;
 }
+
 // ========= 服务方法参数类型 =========
+
+
+
 // ========= 工具定义辅助函数 =========
+
 function makeStringProp(description: string): ToolParamProperty {
    const result: ToolParamProperty = { type: 'string', description: description };
    return result;
 }
+
 function makeIntegerProp(description: string): ToolParamProperty {
    const result: ToolParamProperty = { type: 'integer', description: description };
    return result;
 }
+
 function makeObjectProp(description: string, props: ToolPropertiesDefinition, required?: string[]): ToolParamProperty {
    const param: ToolParamProperty = { type: 'object', description: description };
    param.properties = props;
@ -145,22 +181,27 @@ function makeObjectProp(description: string, props: ToolPropertiesDefinition, re
    }
    return param;
 }
+
 function makeArrayProp(description: string, item: ToolParamProperty): ToolParamProperty {
    const result: ToolParamProperty = { type: 'array', description: description, items: item };
    return result;
 }
+
 function makeBoolProp(description: string): ToolParamProperty {
    const result: ToolParamProperty = { type: 'boolean', description: description };
    return result;
 }
+
 function makeNumberProp(description: string): ToolParamProperty {
    const result: ToolParamProperty = { type: 'number', description: description };
    return result;
 }
+
 function makeEnumProp(description: string, enumValues: string[]): ToolParamProperty {
    const result: ToolParamProperty = { type: 'string', description: description, enum: enumValues };
    return result;
 }
+
 function makeToolDef(name: string, description: string, properties: ToolPropertiesDefinition, required?: string[]): ToolFunctionDef {
    const params: ToolParamDecl = { type: 'object', properties: properties };
    const func: ToolFunctionDecl = { name: name, description: description, parameters: params };
@ -170,6 +211,7 @@ function makeToolDef(name: string, description: string, properties: ToolProperti
    }
    return tool;
 }
+
 /**
 * 系统提示词 — AI 人设 + 图记忆使用说明
 */
@ -232,7 +274,9 @@ ${personaContext || '你是一个帮助用户记录和回忆信息的助手。

 推理得到的信息可以写入但需标注 [推测]。`;
 }
+
 // ========= 工具定义 =========
+
 // Pre-typed property dictionaries for tool definitions
 const recallTimeRangeDict: ToolPropertiesDefinition = { days: makeIntegerProp('最近N天') };
 const tripletPropsDict: ToolPropertiesDefinition = {
@ -255,6 +299,7 @@ const newRelDict: ToolPropertiesDefinition = {
    target: makeStringProp('')
 };
 const EMPTY_PROPS: ToolPropertiesDefinition = {};
+
 const recallProps: ToolPropertiesDefinition = {
    queryIntent: makeStringProp('查询意图，支持逗号分隔多个关键词'),
    seedEntities: makeArrayProp('种子实体（可选）', makeStringProp('')),
@ -324,6 +369,7 @@ const contextRewriteProps: ToolPropertiesDefinition = {
 const personaRemoveProps: ToolPropertiesDefinition = {
    attribute: makeStringProp('要删除的属性名（如：扮演角色、说话风格）')
 };
+
 const TOOLS_DEFINITION: ToolFunctionDef[] = [
    makeToolDef('memory_recall', '检索记忆。支持关键词、时间范围、会话过滤。返回相关实体和关系。\n\n【⚠️ 强制执行顺序 - 每轮必须严格遵守】\n1. 步骤1（必须首先执行）: 查询人设图\n2. 步骤2（必须第二步执行）: 查询工作记忆链\n【重要】跳过步骤1或步骤2将导致系统错误！', recallProps, ['queryIntent']),
    makeToolDef('memory_commit', '写入记忆。将三元组写入图数据库，支持批量写入。\n\n【重要】写入原则:\n- 用户明确表达的信息 → 必须写入\n- AI推理得到的信息 → 可以写入，但需标注[推测]\n- 避免写入冗余或无意义的信息', commitProps, ['triplets']),
@ -343,10 +389,36 @@ const TOOLS_DEFINITION: ToolFunctionDef[] = [
    makeToolDef('task_archive', '归档已完成/过期的任务。将任务状态设为 archived，同时写入完成摘要到图数据库。\n\n【使用场景】\n1. 话题转变时归档旧任务\n2. 已完成的任务及时归档\n3. 长时间无更新的任务归档\n\n【注意】优先使用 task_archive 替代 task_set_state(state=archived)，因为它会自动写入完成摘要。', archiveProps, ['taskId']),
    makeToolDef('task_query', '查询最近的任务列表。按更新时间倒序排列。新对话开始时优先使用此工具获取所有进展中的任务，避免重复创建。', queryProps)
 ];
+
 // ========= 工具名称映射 =========
+
 // Types for executeTool generic args
+
+
+
+
+
 type ToolStateArg = '进行中' | '已完成' | '已暂停' | '已取消';
-type ToolHandlerName = 'memoryRecal' | 'memoryCommit' | 'memoryPurge' | 'memoryIntrospect' | 'memoryArchive' | 'memoryCleanup' | 'memoryQueryArchived' | 'contextRewrite' | 'personaUpdate' | 'personaRemove' | 'personaClear' | 'taskCreate' | 'taskSetState' | 'taskDelete' | 'taskLinkInfo' | 'taskArchive' | 'taskQuery';
+
+type ToolHandlerName =
+    | 'memoryRecal'
+    | 'memoryCommit'
+    | 'memoryPurge'
+    | 'memoryIntrospect'
+    | 'memoryArchive'
+    | 'memoryCleanup'
+    | 'memoryQueryArchived'
+    | 'contextRewrite'
+    | 'personaUpdate'
+    | 'personaRemove'
+    | 'personaClear'
+    | 'taskCreate'
+    | 'taskSetState'
+    | 'taskDelete'
+    | 'taskLinkInfo'
+    | 'taskArchive'
+    | 'taskQuery';
+
 const TOOL_HANDLER_MAP: Record<string, ToolHandlerName> = {
    'memory_recall': 'memoryRecal',
    'memory_commit': 'memoryCommit',
@ -366,20 +438,26 @@ const TOOL_HANDLER_MAP: Record<string, ToolHandlerName> = {
    'task_archive': 'taskArchive',
    'task_query': 'taskQuery',
 };
+
 // ========= AIAgentService =========
+
 export class AIAgentService {
    private memoryService: GraphMemoryService;
    private currentSessionId: string;
    private turnCounter: number = 0;
-    private appContext: common.Context;
-    constructor(memoryService: GraphMemoryService, appContext: common.Context, sessionId?: string) {
+
+    private appContext: Context;
+
+    constructor(memoryService: GraphMemoryService, appContext: Context, sessionId?: string) {
        this.memoryService = memoryService;
        this.appContext = appContext;
        this.currentSessionId = sessionId || `session-hm-${Date.now()}`;
    }
+
    getSessionId(): string {
        return this.currentSessionId;
    }
+
    /**
     * 发送消息 — 完整的 Agent 流程
     * 1. 查询人设
@ -390,14 +468,17 @@ export class AIAgentService {
     */
    async sendMessage(userInput: string): Promise<AgentResponse> {
        this.turnCounter++;
+
        // === 步骤1+2: 获取上下文 ===
        const personaResult = await this.memoryService.personaQuery();
        const personaContext: string = personaResult.found ? this.formatPersona(personaResult.persona) : '';
+
        const recallParams: MemoryRecallParams = {
            queryIntent: 'TaskNode,工作记忆,任务链',
            depth: 2
        };
        const taskResult = await this.memoryService.memoryRecall(recallParams);
+
        // === 读取 API 配置 ===
        const context = this.appContext;
        const pref = await dataPreferences.getPreferences(context, 'trulymem_config');
@ -411,43 +492,33 @@ export class AIAgentService {
            };
            return noKeyResponse;
        }
+
        // === 构建上下文丰富的消息 ===
        const systemPrompt: string = buildSystemPrompt(personaContext);
        const contextBlock: string = this.buildContextBlock(personaResult, taskResult);
        const sysMsg: ApiRequestMessage = { role: 'system' as string, content: systemPrompt };
        const userMsg: ApiRequestMessage = { role: 'user' as string, content: contextBlock + '\n\n---\n\n用户消息: ' + userInput };
        const messages: ApiRequestMessage[] = [sysMsg, userMsg];
+
        // === 步骤3: 请求 AI ===
        const response: ApiResponse = await this.callApi(messages, baseUrl, model, apiKey);
+
        const toolCalls: ToolCallResult[] = [];
+
        // === 步骤4: 处理 tool_calls ===
        if (response.choices && response.choices.length > 0) {
            const choice: ApiChoice = response.choices[0];
            const aiMessage: ApiChoiceMessage = choice.message;
+
            // 处理函数调用
            if (aiMessage.tool_calls && aiMessage.tool_calls.length > 0) {
                for (const tc of aiMessage.tool_calls) {
                    const handlerName: ToolHandlerName | undefined = TOOL_HANDLER_MAP[tc.function.name];
                    if (handlerName) {
-                        let args: Record<string, Object> = {};
-                        try {
-                            args = JSON.parse(tc.function.arguments);
-                        }
-                        catch (parseErr) {
-                            const parseErrMsg = (parseErr as Error).message || JSON.stringify(parseErr);
-                            defaultLogger.error('Failed to parse tool arguments: ' + parseErrMsg);
-                            const badArgsResult: ToolCallResult = {
-                                name: tc.function.name,
-                                success: false,
-                                message: '参数解析失败'
-                            };
-                            toolCalls.push(badArgsResult);
-                            continue;
-                        }
+                        const args: Record<string, Object> = JSON.parse(tc.function.arguments);
                        const result: ToolCallResult = await this.executeTool(handlerName, args);
                        toolCalls.push(result);
-                    }
-                    else {
+                    } else {
                        const unknownToolResult: ToolCallResult = {
                            name: tc.function.name,
                            success: false,
@ -456,6 +527,7 @@ export class AIAgentService {
                        toolCalls.push(unknownToolResult);
                    }
                }
+
                // 有 tool_calls 时需要再次请求 AI，带上工具执行结果
                const followUpSystemMsg: ApiRequestMessage = { role: 'system', content: systemPrompt };
                const followUpUserMsg: ApiRequestMessage = { role: 'user', content: contextBlock + '\n\n---\n\n用户消息: ' + userInput };
@ -468,6 +540,7 @@ export class AIAgentService {
                    followUpUserMsg,
                    followUpAssistantMsg,
                ];
+
                for (const tc of aiMessage.tool_calls) {
                    const callResult: ToolCallResult | undefined = toolCalls.find(r => r.name === tc.function.name);
                    const toolResultMsg: string = callResult ? callResult.message : '完成';
@ -477,6 +550,7 @@ export class AIAgentService {
                    };
                    toolResultsMessages.push(toolResultMessage);
                }
+
                const finalResponse: ApiResponse = await this.callApi(toolResultsMessages, baseUrl, model, apiKey);
                if (finalResponse.choices && finalResponse.choices.length > 0) {
                    const content: string = finalResponse.choices[0].message.content || '';
@ -484,21 +558,29 @@ export class AIAgentService {
                    return finalResult;
                }
            }
+
            // 普通回复（无 tool_calls）
            const content: string = aiMessage.content || '';
            const noToolResponse: AgentResponse = { content, toolCalls };
            return noToolResponse;
        }
+
        const noResponse: AgentResponse = {
            content: 'AI 无响应',
            toolCalls
        };
        return noResponse;
    }
+
    /**
     * 请求 DeepSeek API
     */
-    private async callApi(messages: ApiRequestMessage[], baseUrl: string, model: string, apiKey: string): Promise<ApiResponse> {
+    private async callApi(
+        messages: ApiRequestMessage[],
+        baseUrl: string,
+        model: string,
+        apiKey: string
+    ): Promise<ApiResponse> {
        const httpRequest = http.createHttp();
        try {
            const resp = await httpRequest.request(baseUrl + '/chat/completions', {
@ -521,11 +603,11 @@ export class AIAgentService {
            }
            const errorMsg: string = `API 请求失败: HTTP ${resp.responseCode}`;
            throw new Error(errorMsg);
-        }
-        finally {
+        } finally {
            httpRequest.destroy();
        }
    }
+
    /**
     * 执行工具调用
     */
@ -548,6 +630,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryCommit': {
                    const commitParams: MemoryCommitParams = {
                        triplets: args.triplets as TripletInput[],
@ -563,6 +646,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryPurge': {
                    const purgeArgs: MemoryPurgeParams = {
                        criteria: args.criteria as PurgeCriteriaParams,
@ -577,6 +661,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryIntrospect': {
                    const introspectResult = await this.memoryService.memoryIntrospect(args.sessionId as string);
                    const result: ToolCallResult = {
@ -586,6 +671,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryArchive': {
                    const archiveResult = await this.memoryService.archive(args.days as number);
                    const result: ToolCallResult = {
@ -595,6 +681,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryCleanup': {
                    const cleanupResult = await this.memoryService.cleanup((args.dryRun as boolean) !== false);
                    const result: ToolCallResult = {
@ -604,6 +691,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'memoryQueryArchived': {
                    const qaResult = await this.memoryService.queryArchived(args.days as number, args.keyword as string);
                    const result: ToolCallResult = {
@ -613,6 +701,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'contextRewrite': {
                    const summary = args.summary as string;
                    const result: ToolCallResult = {
@ -622,6 +711,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'personaUpdate': {
                    const personaParams: PersonaUpdateParams = {
                        tone: args.tone as string,
@ -638,6 +728,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'personaRemove': {
                    const prResult = await this.memoryService.personaRemove(args.attribute as string);
                    const result: ToolCallResult = {
@ -647,6 +738,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'personaClear': {
                    const pcResult = await this.memoryService.personaClear();
                    const result: ToolCallResult = {
@ -656,6 +748,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskCreate': {
                    const createParams: TaskCreateParams = {
                        taskId: args.taskId as string,
@ -670,6 +763,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskSetState': {
                    const setStateParams: TaskSetStateParams = {
                        taskId: args.taskId as string,
@ -683,6 +777,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskDelete': {
                    const deleteParams: TaskDeleteParams = {
                        taskId: args.taskId as string,
@ -696,6 +791,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskLinkInfo': {
                    const linkInfoParams: TaskLinkInfoParams = {
                        taskId: args.taskId as string,
@ -709,6 +805,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskArchive': {
                    const archiveParams: TaskArchiveParams = {
                        taskId: args.taskId as string,
@ -722,6 +819,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                case 'taskQuery': {
                    const tqResult = await this.memoryService.taskQuery({
                        limit: args.limit as number,
@ -734,6 +832,7 @@ export class AIAgentService {
                    };
                    return result;
                }
+
                default: {
                    const defaultResult: ToolCallResult = {
                        name: name as string,
@ -743,8 +842,7 @@ export class AIAgentService {
                    return defaultResult;
                }
            }
-        }
-        catch (e) {
+        } catch (e) {
            const errorMessage: string = (e as Error).message || '';
            const errorResult: ToolCallResult = {
                name: name as string,
@ -754,6 +852,7 @@ export class AIAgentService {
            return errorResult;
        }
    }
+
    /**
     * 格式化人设数据为文本
     */
@ -767,19 +866,26 @@ export class AIAgentService {
        }
        return parts.length > 0 ? parts.join('；') : '';
    }
+
    /**
     * 构建上下文注入块
     */
-    private buildContextBlock(personaResult: PersonaQueryResult, taskResult: MemoryRecallResult): string {
+    private buildContextBlock(
+        personaResult: PersonaQueryResult,
+        taskResult: MemoryRecallResult
+    ): string {
        const blocks: string[] = [];
+
        if (personaResult.found) {
            blocks.push(`【当前人设】\n${this.formatPersona(personaResult.persona)}`);
        }
+
        if (taskResult.entities.length > 0) {
            const entitySample: EntityInfo[] = taskResult.entities.slice(0, 5);
            const entitiesStr: string = JSON.stringify(entitySample);
            blocks.push(`【工作记忆】\n${taskResult.message}\n${entitiesStr}`);
        }
+
        return blocks.length > 0 ? blocks.join('\n\n') : '【新对话】';
    }
 }
--- a/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/service/GraphMemoryService.ts
+++ b/products/phone/build/default/cache/default/default@CompileArkTS/esmodule/debug/common/src/main/ets/service/GraphMemoryService.ts
@ -1,10 +1,18 @@
-import type { GraphDatabase, RelationQueryResult, TimeRangeParams } from '../model/GraphDatabase';
-import { defaultLogger } from "@normalized:N&&&@ohos/common/src/main/ets/util/Logger&1.0.0";
+/**
+ * GraphMemoryService - 图记忆服务层
+ * 封装 GraphDatabase，提供 AI Agent 友好的图记忆操作方法
+ * 参考：main 分支 core/tools/memory_tools.py
+ */
+import { GraphDatabase, RelationQueryResult, TimeRangeParams } from '../model/GraphDatabase';
+import { defaultLogger } from '../util/Logger';
+
 // ========= 接口定义 =========
+
 export interface SnapshotData {
    entities: EntityInfo[];
    relations: RelationInfo[];
 }
+
 export interface GraphDataNode {
    id: number;
    label: string;
@ -12,6 +20,7 @@ export interface GraphDataNode {
    mentions: number;
    depth?: number;
 }
+
 export interface GraphDataEdge {
    from: number;
    to: number;
@ -21,12 +30,14 @@ export interface GraphDataEdge {
    sessionId?: string;
    turnId?: number;
 }
+
 export interface SnapshotEntity {
    name: string;
    type: string;
    mention_count: number;
    depth?: number;
 }
+
 export interface SnapshotRelation {
    source: string;
    target: string;
@ -36,20 +47,24 @@ export interface SnapshotRelation {
    turn_id?: number;
    depth?: number;
 }
+
 export interface GraphOutput {
    nodes: GraphDataNode[];
    edges: GraphDataEdge[];
 }
+
 export interface SnapshotOutput {
    entities: SnapshotEntity[];
    relations: SnapshotRelation[];
 }
+
 export interface EntityInfo {
    name: string;
    type: string;
    mentionCount: number;
    depth?: number;
 }
+
 export interface RelationInfo {
    source: string;
    target: string;
@ -59,12 +74,14 @@ export interface RelationInfo {
    turnId?: number;
    depth?: number;
 }
+
 export interface TripletInput {
    subject: string;
    relation: string;
    object: string;
    confidence?: number;
 }
+
 export interface CleanupResult {
    cleaned: number;
    deletedRelations?: number;
@ -72,11 +89,13 @@ export interface CleanupResult {
    dryRun?: boolean;
    message: string;
 }
+
 export interface SearchResult {
    name: string;
    type: string;
    mentions: number;
 }
+
 export interface TaskInfo {
    taskId: string;
    description: string;
@ -84,23 +103,28 @@ export interface TaskInfo {
    infoCount: number;
    updatedAt: string;
 }
+
 export interface NodeData {
    id: number;
    label: string;
    type: string;
    mentions: number;
 }
+
 export interface EdgeData {
    from: number;
    to: number;
    label: string;
    weight: number;
 }
+
 export interface GraphData {
    nodes: NodeData[];
    edges: EdgeData[];
 }
+
 // 内部接口 — 用于替换内联对象类型声明
+
 export interface PurgeCriteriaParams {
    subjectContains?: string;
    relationType?: string;
@ -110,26 +134,31 @@ export interface PurgeCriteriaParams {
    targetType?: string;
    sourceHasStatus?: string;
 }
+
 export interface NewRelationParams {
    relation: string;
    target: string;
 }
+
 interface TripletData {
    subject: string;
    relation: string;
    object: string;
 }
+
 interface CriteriaData {
    subject?: string;
    target?: string;
    relation?: string;
    sessionId?: string;
 }
+
 export interface MemoryRecallResult {
    entities: EntityInfo[];
    relations: RelationInfo[];
    message: string;
 }
+
 export interface MemoryRecallParams {
    queryIntent: string;
    seedEntities?: string[];
@ -137,61 +166,74 @@ export interface MemoryRecallParams {
    timeRange?: TimeRangeParams;
    sessionFilter?: string;
 }
+
 export interface MemoryCommitParams {
    triplets: TripletInput[];
    entityTypes?: Record<string, string>;
    sessionId?: string;
    turnId?: number;
 }
+
 interface MemoryCommitResult {
    committedCount: number;
    details: string[];
 }
+
 export interface MemoryPurgeParams {
    criteria: PurgeCriteriaParams;
    mode: 'soft' | 'hard' | 'supersede';
    newRelation?: NewRelationParams;
 }
+
 interface MemoryPurgeResult {
    deletedCount: number;
    message: string;
 }
+
 interface HotNodeInfo {
    name: string;
    mentionCount: number;
    type: string;
 }
+
 interface MemoryIntrospectResult {
    entityCount: number;
    relationCount: number;
    hotNodes: HotNodeInfo[];
    message: string;
 }
+
 interface ArchiveResult {
    archived: number;
    message: string;
 }
+
 interface PersonaResult {
    success: boolean;
    message: string;
 }
+
 export interface PersonaQueryResult {
    persona: Record<string, string>;
    found: boolean;
 }
+
 interface TaskCreateResult {
    success: boolean;
    taskId: string;
    message: string;
 }
+
 interface TaskActionResult {
    success: boolean;
    message: string;
 }
+
 export interface TaskQueryResult {
    tasks: TaskInfo[];
    message: string;
 }
+
 export interface PersonaUpdateParams {
    tone?: string;
    style?: string;
@ -199,36 +241,44 @@ export interface PersonaUpdateParams {
    catchphrase?: string;
    background?: string;
 }
+
 export interface TaskCreateParams {
    taskId: string;
    description: string;
    infoNodes?: string[];
 }
+
 export interface TaskSetStateParams {
    taskId: string;
-    state: '进行中' | '已完成' | '已暂停' | '已取消' | '已归档';
+    state: '进行中' | '已完成' | '已暂停' | '已取消';
 }
+
 export interface TaskDeleteParams {
    taskId: string;
    deleteInfoNodes?: boolean;
 }
+
 export interface TaskLinkInfoParams {
    taskId: string;
    infoNodeNames: string[];
 }
+
 export interface TaskArchiveParams {
    taskId: string;
    summary?: string;
 }
+
 export interface TaskQueryParams {
    limit?: number;
    stateFilter?: string;
 }
+
 // 节点详情连接项接口
 export interface ConnectionItem {
    type: string;
    target_name: string;
 }
+
 // 节点详情返回接口
 export interface NodeDetailInfo {
    name: string;
@ -237,28 +287,41 @@ export interface NodeDetailInfo {
    connection_count: number;
    connections: ConnectionItem[];
 }
+
 // 图数据统计接口
 export interface GraphStats {
    maxDegree: number;
    avgDegree: number;
 }
+
 // 人设节点的固定名称
 const PERSONA_NODE_NAME: string = 'trulymem_persona_identity';
 const PERSONA_NODE_TYPE: string = 'PersonaNode';
+
 // ========= GraphMemoryService =========
+
 export class GraphMemoryService {
    private db: GraphDatabase;
+
    constructor(db: GraphDatabase) {
        this.db = db;
    }
+
    // ========= 记忆操作 =========
+
    /**
     * 记忆召回 — 关键词搜索 + BFS 扩展
     * 对应 tools.memory_recall
     */
    async memoryRecall(params: MemoryRecallParams): Promise<MemoryRecallResult> {
        const depth: number = params.depth ?? 2;
-        const result = await this.db.recall(params.queryIntent, params.seedEntities, depth, undefined, params.sessionFilter);
+        const result = await this.db.recall(
+            params.queryIntent,
+            params.seedEntities,
+            depth,
+            undefined,
+            params.sessionFilter
+        );
        const entities: EntityInfo[] = result.entities.map(e => {
            const entityItem: EntityInfo = {
                name: e.name,
@ -282,6 +345,7 @@ export class GraphMemoryService {
        });
        return { entities, relations, message: result.message };
    }
+
    /**
     * 记忆写入 — 批量三元组
     * 对应 tools.memory_commit
@ -296,7 +360,12 @@ export class GraphMemoryService {
                continue;
            }
            const triplets: TripletData[] = [{ subject, relation, object }];
-            await this.db.commit(triplets, params.entityTypes, params.sessionId, params.turnId);
+            await this.db.commit(
+                triplets,
+                params.entityTypes,
+                params.sessionId,
+                params.turnId
+            );
            details.push(`${subject} -[${relation}]-> ${object}`);
        }
        const result: MemoryCommitResult = {
@ -305,6 +374,7 @@ export class GraphMemoryService {
        };
        return result;
    }
+
    /**
     * 记忆删除
     * 对应 tools.memory_purge
@ -321,14 +391,15 @@ export class GraphMemoryService {
                relation: softRelation,
                target: softTarget
            }, 'soft');
+
            // 新建替代关系
            const newSubj = softSubject || '';
            if (newSubj) {
                await this.db.commit([{
-                        subject: newSubj,
-                        relation: params.newRelation.relation,
-                        object: params.newRelation.target
-                    }]);
+                    subject: newSubj,
+                    relation: params.newRelation.relation,
+                    object: params.newRelation.target
+                }]);
            }
            const supersedeResult: MemoryPurgeResult = {
                deletedCount: 1,
@ -336,6 +407,7 @@ export class GraphMemoryService {
            };
            return supersedeResult;
        }
+
        const subjContains = params.criteria.subjectContains;
        const relType = params.criteria.relationType;
        const tgtContains = params.criteria.targetContains;
@ -357,6 +429,7 @@ export class GraphMemoryService {
        };
        return purgeResult;
    }
+
    /**
     * 记忆状态查询
     */
@ -382,18 +455,21 @@ export class GraphMemoryService {
        };
        return result;
    }
+
    /**
     * 关键词搜索节点
     */
    async search(keyword: string): Promise<SearchResult[]> {
        return await this.db.search(keyword);
    }
+
    /**
     * 归档旧记忆
     */
    async archive(days: number): Promise<ArchiveResult> {
        return await this.db.archive(days);
    }
+
    /**
     * 清理已删除的记忆
     */
@ -408,6 +484,7 @@ export class GraphMemoryService {
        };
        return cleanupResult;
    }
+
    /**
     * 记忆图谱 — 在指定时间范围内查询关系
     * 对应 tools.memory_graph
@ -464,6 +541,7 @@ export class GraphMemoryService {
        };
        return out;
    }
+
    /**
     * 记忆快照 — 在指定时间范围内查询实体和关系
     * 对应 tools.memory_snapshot
@ -518,6 +596,7 @@ export class GraphMemoryService {
        };
        return out2;
    }
+
    /**
     * 记忆清理 — 归档旧记忆并清理孤立节点
     * 对应 tools.memory_cleanup
@ -533,10 +612,13 @@ export class GraphMemoryService {
        };
        return cleanupResult;
    }
+
    /**
     * 记忆清理 — 删除指定条件的记忆
     * 对应 tools.memory_purge
     */
+
+
    /**
     * 查询已归档的记忆
     * 对应 tools.memory_query_archived
@ -545,13 +627,14 @@ export class GraphMemoryService {
        try {
            const result = await this.db.queryArchived(days, keyword);
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            defaultLogger.error('queryArchived error: ' + JSON.stringify(e));
            return [];
        }
    }
+
    // ========= 人设管理 =========
+
    /**
     * 更新人设
     * 对应 tools.persona_update
@ -563,12 +646,14 @@ export class GraphMemoryService {
            const personaTriplets: TripletData[] = [];
            const entityTypes: Record<string, string> = {};
            entityTypes[PERSONA_NODE_NAME] = PERSONA_NODE_TYPE;
+
            // 2. 逐个属性写入（作为关系），不使用 as any
            const toneVal = params.tone;
            const styleVal = params.style;
            const personalityVal = params.personality;
            const catchphraseVal = params.catchphrase;
            const backgroundVal = params.background;
+
            if (toneVal) {
                personaTriplets.push({
                    subject: PERSONA_NODE_NAME,
@ -604,6 +689,7 @@ export class GraphMemoryService {
                    object: backgroundVal
                });
            }
+
            if (personaTriplets.length > 0) {
                await this.db.commit(personaTriplets, entityTypes);
            }
@ -612,8 +698,7 @@ export class GraphMemoryService {
                message: `已更新 ${personaTriplets.length} 个人设属性`
            };
            return successResult;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: PersonaResult = {
                success: false,
                message: `更新人设失败: ${(e as Error).message || ''}`
@ -621,6 +706,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 清除人设
     * 对应 tools.persona_clear
@ -630,8 +716,7 @@ export class GraphMemoryService {
            await this.db.purge({ subject: PERSONA_NODE_NAME }, 'hard');
            const result: PersonaResult = { success: true, message: '已清除所有人设信息' };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: PersonaResult = {
                success: false,
                message: `清除人设失败: ${(e as Error).message || ''}`
@ -639,6 +724,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 删除单条人设属性
     * 对应 tools.persona_remove
@ -652,14 +738,14 @@ export class GraphMemoryService {
                success: true,
                message: `已删除人设属性: ${attribute}`
            };
-        }
-        catch (e) {
+        } catch (e) {
            return {
                success: false,
                message: `删除人设属性失败: ${(e as Error).message || ''}`
            };
        }
    }
+
    /**
     * 查询当前人设
     */
@ -678,7 +764,9 @@ export class GraphMemoryService {
        };
        return queryResult;
    }
+
    // ========= 任务管理 =========
+
    /**
     * 创建任务节点
     * 对应 tools.task_create
@ -687,16 +775,19 @@ export class GraphMemoryService {
        try {
            const entityTypes: Record<string, string> = {};
            entityTypes[params.taskId] = 'TaskNode';
+
            const triplets: TripletData[] = [
                { subject: params.taskId, relation: 'description', object: params.description },
                { subject: params.taskId, relation: 'has_state', object: '进行中' }
            ];
+
            if (params.infoNodes && params.infoNodes.length > 0) {
                for (const infoNode of params.infoNodes) {
                    entityTypes[infoNode] = 'InfoNode';
                    triplets.push({ subject: params.taskId, relation: 'CONTAINS_INFO', object: infoNode });
                }
            }
+
            await this.db.commit(triplets, entityTypes);
            const result: TaskCreateResult = {
                success: true,
@ -704,8 +795,7 @@ export class GraphMemoryService {
                message: `已创建任务: ${params.taskId}`
            };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: TaskCreateResult = {
                success: false,
                taskId: params.taskId,
@ -714,6 +804,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 设置任务状态
     * 对应 tools.task_set_state
@ -728,8 +819,7 @@ export class GraphMemoryService {
                message: `任务 ${params.taskId} 状态已设为: ${params.state}`
            };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: TaskActionResult = {
                success: false,
                message: `设置任务状态失败: ${(e as Error).message || ''}`
@ -737,6 +827,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 删除任务
     * 对应 tools.task_delete
@ -750,8 +841,7 @@ export class GraphMemoryService {
            }
            const result: TaskActionResult = { success: true, message: `已删除任务: ${params.taskId}` };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: TaskActionResult = {
                success: false,
                message: `删除任务失败: ${(e as Error).message || ''}`
@ -759,6 +849,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 关联信息节点到任务
     * 对应 tools.task_link_info
@ -774,8 +865,7 @@ export class GraphMemoryService {
            await this.db.commit(triplets, entityTypes);
            const result: TaskActionResult = { success: true, message: `已关联 ${triplets.length} 个信息节点` };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: TaskActionResult = {
                success: false,
                message: `关联信息节点失败: ${(e as Error).message || ''}`
@ -783,14 +873,14 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 归档任务
     * 对应 tools.task_archive
     */
    async taskArchive(params: TaskArchiveParams): Promise<TaskActionResult> {
        try {
-            // 归档任务：将状态设为已归档（archived），同时写入完成摘要
-            await this.taskSetState({ taskId: params.taskId, state: '已归档' });
+            await this.taskSetState({ taskId: params.taskId, state: '已暂停' });
            if (params.summary) {
                await this.db.commit([
                    { subject: params.taskId, relation: 'archive_summary', object: params.summary }
@ -798,8 +888,7 @@ export class GraphMemoryService {
            }
            const result: TaskActionResult = { success: true, message: `已归档任务: ${params.taskId}` };
            return result;
-        }
-        catch (e) {
+        } catch (e) {
            const errorResult: TaskActionResult = {
                success: false,
                message: `归档任务失败: ${(e as Error).message || ''}`
@ -807,6 +896,7 @@ export class GraphMemoryService {
            return errorResult;
        }
    }
+
    /**
     * 查询任务列表
     * 对应 tools.task_query
@ -820,7 +910,9 @@ export class GraphMemoryService {
        };
        return queryResult;
    }
+
    // ========= 图数据 =========
+
    /**
     * 获取用于 WebView 的完整图数据
     */
@ -831,6 +923,7 @@ export class GraphMemoryService {
        const edges: EdgeData[] = [];
        let nodeIdCounter = 1;
        const nameToId: Record<string, number> = {};
+
        for (const entity of recallResult.entities) {
            const id = nodeIdCounter;
            nodeIdCounter++;
@ -843,6 +936,7 @@ export class GraphMemoryService {
            };
            nodes.push(node);
        }
+
        for (const rel of recallResult.relations) {
            const from = nameToId[rel.source];
            const to = nameToId[rel.target];
@ -856,9 +950,11 @@ export class GraphMemoryService {
                edges.push(edge);
            }
        }
+
        const graphData: GraphData = { nodes, edges };
        return graphData;
    }
+
    /**
     * 查询节点的完整信息 — 自身属性 + 所有相连关系
     */
@ -868,9 +964,11 @@ export class GraphMemoryService {
            if (recallResult.entities.length === 0) {
                return null;
            }
+
            const entity = recallResult.entities[0];
            const connections: ConnectionItem[] = [];
            let connectionCount = 0;
+
            for (const rel of recallResult.relations) {
                if (rel.source === nodeName) {
                    const conn: ConnectionItem = {
@ -879,8 +977,7 @@ export class GraphMemoryService {
                    };
                    connections.push(conn);
                    connectionCount++;
-                }
-                else if (rel.target === nodeName) {
+                } else if (rel.target === nodeName) {
                    const conn: ConnectionItem = {
                        type: rel.type + ' (反向)',
                        target_name: rel.source
@ -889,6 +986,7 @@ export class GraphMemoryService {
                    connectionCount++;
                }
            }
+
            const detail: NodeDetailInfo = {
                name: entity.name,
                type: entity.type,
@ -897,23 +995,25 @@ export class GraphMemoryService {
                connections
            };
            return detail;
-        }
-        catch (err) {
+        } catch (err) {
            defaultLogger.error('getNodeDetail error: ' + JSON.stringify(err));
            return null;
        }
    }
+
    /**
     * 返回带连接度数的图数据（每个节点增加 degree 字段）
     */
    async getJoinedData(): Promise<GraphData> {
        const graphData = await this.getGraphDataForView();
+
        // 计算每个节点的连接度数
        const degreeMap: Record<number, number> = {};
        for (const edge of graphData.edges) {
            degreeMap[edge.from] = (degreeMap[edge.from] || 0) + 1;
            degreeMap[edge.to] = (degreeMap[edge.to] || 0) + 1;
        }
+
        // 手动为节点附加 degree（ArkTS 不支持展开运算符）
        const nodesWithDegree: NodeData[] = [];
        for (let i = 0; i < graphData.nodes.length; i++) {
@ -926,6 +1026,7 @@ export class GraphMemoryService {
            };
            nodesWithDegree.push(copy);
        }
+
        const graphResult: GraphData = {
            nodes: nodesWithDegree,
            edges: graphData.edges
--- a/common/src/main/ets/util/BreakpointSystem.ets
+++ b/common/src/main/ets/util/BreakpointSystem.ets
@ -0,0 +1,47 @@
+export enum WidthBreakpoint {
+  WIDTH_XS = 'xs',
+  WIDTH_SM = 'sm',
+  WIDTH_MD = 'md',
+  WIDTH_LG = 'lg',
+  WIDTH_XL = 'xl'
+}
+
+export interface BreakpointTypes<T> {
+  xs?: T;
+  sm: T;
+  md: T;
+  lg: T;
+  xl?: T;
+}
+
+export class BreakpointType<T> {
+  private xs: T;
+  private sm: T;
+  private md: T;
+  private lg: T;
+  private xl: T;
+
+  public constructor(param: BreakpointTypes<T>) {
+    this.xs = param.xs ?? param.sm;
+    this.sm = param.sm;
+    this.md = param.md;
+    this.lg = param.lg;
+    this.xl = param.xl ?? param.lg;
+  }
+
+  public getValue(currentBreakpoint: WidthBreakpoint): T {
+    if (currentBreakpoint === WidthBreakpoint.WIDTH_XS) {
+      return this.xs;
+    }
+    if (currentBreakpoint === WidthBreakpoint.WIDTH_SM) {
+      return this.sm;
+    }
+    if (currentBreakpoint === WidthBreakpoint.WIDTH_MD) {
+      return this.md;
+    }
+    if (currentBreakpoint === WidthBreakpoint.WIDTH_XL) {
+      return this.xl;
+    }
+    return this.lg;
+  }
+}
--- a/common/src/main/ets/util/Logger.ets
+++ b/common/src/main/ets/util/Logger.ets
@ -0,0 +1,31 @@
+import { hilog } from "@kit.PerformanceAnalysisKit";
+
+class Logger {
+  private domain: number;
+  private prefix: string;
+  private format: string = "%{public}s, %{public}s";
+
+  public constructor(prefix: string) {
+    this.prefix = prefix;
+    this.domain = 0xFF00;
+  }
+
+  public debug(...args: Object[]): void {
+    hilog.debug(this.domain, this.prefix, this.format, args);
+  }
+
+  public info(...args: Object[]): void {
+    hilog.info(this.domain, this.prefix, this.format, args);
+  }
+
+  public warn(...args: Object[]): void {
+    hilog.warn(this.domain, this.prefix, this.format, args);
+  }
+
+  public error(...args: Object[]): void {
+    hilog.error(this.domain, this.prefix, this.format, args);
+  }
+}
+
+export const defaultLogger = new Logger("[TrulyMEM]");
+export default defaultLogger;
--- a/common/src/main/ets/viewmodel/BaseViewModel.ets
+++ b/common/src/main/ets/viewmodel/BaseViewModel.ets
@ -0,0 +1,52 @@
+import { BreakpointType, WidthBreakpoint } from '../util/BreakpointSystem';
+
+export interface VMEvent {
+}
+
+export class BaseViewModel {
+  protected isAttached: boolean = false;
+  protected isDisposed: boolean = false;
+  protected currentBreakpoint: WidthBreakpoint = WidthBreakpoint.WIDTH_MD;
+
+  attach(): void {
+    if (this.isAttached) {
+      return;
+    }
+    this.isAttached = true;
+    this.onAttach();
+  }
+
+  detach(): void {
+    if (!this.isAttached) {
+      return;
+    }
+    this.isAttached = false;
+    this.onDetach();
+  }
+
+  dispose(): void {
+    if (this.isDisposed) {
+      return;
+    }
+    this.isDisposed = true;
+    this.detach();
+    this.onDispose();
+  }
+
+  protected onAttach(): void {
+  }
+
+  protected onDetach(): void {
+  }
+
+  protected onDispose(): void {
+  }
+
+  public get attached(): boolean {
+    return this.isAttached;
+  }
+
+  public get disposed(): boolean {
+    return this.isDisposed;
+  }
+}
--- a/common/src/main/module.json5
+++ b/common/src/main/module.json5
@ -0,0 +1,12 @@
+{
+  "module": {
+    "name": "common",
+    "type": "har",
+    "description": "TrulyMEM common module",
+    "deviceTypes": [
+      "phone",
+      "tablet",
+      "2in1"
+    ]
+  }
+}
--- a/core/init.py
+++ b/core/init.py
@ -1,12 +0,0 @@
-from .server import BackendServer, Packet, PacketType, PacketResponse
-from .client import BackendClient
-from .embedded_db import EmbeddedGraphDB
-
-__all__ = [
-    "BackendServer",
-    "BackendClient",
-    "EmbeddedGraphDB",
-    "Packet",
-    "PacketType",
-    "PacketResponse"
-]
--- a/core/activity_recorder.py
+++ b/core/activity_recorder.py
@ -1,185 +0,0 @@
-"""
-活动记录器 - 记录 AI 对图数据库的操作
-使用 SQLite :memory: 供 WebUI 实时渲染，同时后台线程持久化到日志文件
-日志每 6 小时自动压缩归档
-"""
-
-import sqlite3
-import time
-import os
-import gzip
-import json
-import threading
-import shutil
-from typing import List, Dict, Optional
-from datetime import datetime, timedelta
-
-
-# 日志目录
-LOG_DIR = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), "logs")
-# 归档间隔（秒）
-ARCHIVE_INTERVAL = 6 * 3600  # 6 小时
-# 轮询间隔（秒）
-POLL_INTERVAL = 10
-
-
-class ActivityRecorder:
-    """记录 AI 对图数据库的操作到内存 SQLite"""
-
-    def __init__(self):
-        self.conn = sqlite3.connect(":memory:", check_same_thread=False)
-        self.conn.execute(
-            "CREATE TABLE activities (id INTEGER PRIMARY KEY AUTOINCREMENT, "
-            "timestamp REAL, action TEXT, tool_name TEXT, entity TEXT, detail TEXT)"
-        )
-        self.conn.commit()
-
-    def record(self, action: str, tool_name: str, entity: str, detail: str = "") -> None:
-        self.conn.execute(
-            "INSERT INTO activities (timestamp, action, tool_name, entity, detail) VALUES (?, ?, ?, ?, ?)",
-            (time.time(), action, tool_name, entity, detail)
-        )
-        self.conn.commit()
-
-    def get_all(self) -> List[Dict]:
-        cursor = self.conn.execute(
-            "SELECT id, timestamp, action, tool_name, entity, detail FROM activities ORDER BY id"
-        )
-        rows = cursor.fetchall()
-        return [
-            {"id": r[0], "timestamp": r[1], "action": r[2],
-             "tool_name": r[3], "entity": r[4], "detail": r[5]}
-            for r in rows
-        ]
-
-    def get_since_id(self, last_id: int) -> List[Dict]:
-        """获取自 last_id 之后的新记录"""
-        cursor = self.conn.execute(
-            "SELECT id, timestamp, action, tool_name, entity, detail FROM activities WHERE id > ? ORDER BY id",
-            (last_id,)
-        )
-        rows = cursor.fetchall()
-        return [
-            {"id": r[0], "timestamp": r[1], "action": r[2],
-             "tool_name": r[3], "entity": r[4], "detail": r[5]}
-            for r in rows
-        ]
-
-    def get_max_id(self) -> int:
-        cursor = self.conn.execute("SELECT COALESCE(MAX(id), 0) FROM activities")
-        return cursor.fetchone()[0]
-
-    def clear(self) -> None:
-        self.conn.execute("DELETE FROM activities")
-        self.conn.commit()
-
-    def get_summary(self) -> Dict[str, int]:
-        cursor = self.conn.execute("SELECT action, COUNT(*) FROM activities GROUP BY action")
-        rows = cursor.fetchall()
-        return {r[0]: r[1] for r in rows}
-
-
-# ── 日志文件管理 ──
-
-def _current_log_path() -> str:
-    """返回当前日志文件路径（按日期命名）"""
-    os.makedirs(LOG_DIR, exist_ok=True)
-    date_str = datetime.now().strftime("%Y%m%d")
-    return os.path.join(LOG_DIR, f"operations.{date_str}.log")
-
-
-def _archive_log(filepath: str) -> str:
-    """压缩归档日志文件，返回归档文件路径"""
-    if not os.path.exists(filepath) or os.path.getsize(filepath) == 0:
-        return ""
-    archive_path = filepath + ".gz"
-    try:
-        with open(filepath, "rb") as f_in:
-            with gzip.open(archive_path, "wb") as f_out:
-                shutil.copyfileobj(f_in, f_out)
-        os.remove(filepath)
-        return archive_path
-    except Exception:
-        return ""
-
-
-class LogPersister:
-    """后台日志持久化线程 - 定期将内存记录写入日志文件并自动归档"""
-
-    def __init__(self, recorder: ActivityRecorder):
-        self.recorder = recorder
-        self._last_persisted_id = 0
-        self._last_archive_time = time.time()
-        self._running = True
-        self._thread = threading.Thread(target=self._run, daemon=True, name="log-persister")
-        self._thread.start()
-
-    def _run(self):
-        """主循环"""
-        while self._running:
-            try:
-                self._persist_new()
-                self._check_archive()
-            except Exception:
-                pass  # 不因日志异常影响主进程
-            time.sleep(POLL_INTERVAL)
-
-    def _persist_new(self):
-        """增量写入新记录到日志文件"""
-        records = self.recorder.get_since_id(self._last_persisted_id)
-        if not records:
-            return
-
-        log_path = _current_log_path()
-        with open(log_path, "a", encoding="utf-8") as f:
-            for r in records:
-                line = json.dumps(r, ensure_ascii=False)
-                f.write(line + "\n")
-
-        # 更新水位
-        if records:
-            self._last_persisted_id = records[-1]["id"]
-
-    def _check_archive(self):
-        """检查是否需要归档"""
-        elapsed = time.time() - self._last_archive_time
-        if elapsed < ARCHIVE_INTERVAL:
-            return
-
-        log_path = _current_log_path()
-        archived = _archive_log(log_path)
-        if archived:
-            dt = datetime.fromtimestamp(self._last_archive_time)
-            print(f"[日志归档] {dt.strftime('%H:%M')} → {os.path.basename(archived)} ({_fmt_size(archived)})")
-        self._last_archive_time = time.time()
-
-    def stop(self):
-        self._running = False
-
-
-def _fmt_size(path: str) -> str:
-    size = os.path.getsize(path)
-    for unit in ("B", "KB", "MB"):
-        if size < 1024:
-            return f"{size:.1f}{unit}"
-        size /= 1024
-    return f"{size:.1f}GB"
-
-
-# ── 单例 ──
-
-_recorder: Optional[ActivityRecorder] = None
-_persister: Optional[LogPersister] = None
-
-
-def get_recorder() -> ActivityRecorder:
-    """获取全局 ActivityRecorder（首次调用时自动启动日志持久化线程）"""
-    global _recorder, _persister
-    if _recorder is None:
-        _recorder = ActivityRecorder()
-        _persister = LogPersister(_recorder)
-    return _recorder
-
-
-def get_persister() -> Optional[LogPersister]:
-    return _persister
--- a/core/client.py
+++ b/core/client.py
@ -1,128 +0,0 @@
-import threading
-import time
-from typing import Any, Dict
-
-from .server import BackendServer, Packet, PacketType
-
-
-class BackendClient:
-    
-    def __init__(self, server: BackendServer):
-        self._server = server
-        self._counter = 0
-        self._lock = threading.Lock()
-
-    def _next_id(self) -> str:
-        with self._lock:
-            self._counter += 1
-            return f"{time.time()}_{self._counter}"
-
-    def send(self, message: str) -> Dict:
-        return self.process_message(message)
-
-    def process_message(self, user_input: str) -> Dict:
-        return self._server.process_message(user_input)
-
-    def get_settings(self) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_SETTINGS,
-            body={}
-        )
-        return self._server.send(packet).body
-
-    def update_settings(self, api_config: Dict = None, tool_limits: Dict = None) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.SET_SETTINGS,
-            body={
-                "api_config": api_config or {},
-                "tool_limits": tool_limits or {}
-            }
-        )
-        return self._server.send(packet).body
-
-    def execute_tool(self, name: str, arguments: Dict) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.EXECUTE_TOOL,
-            body={"tool_name": name, "arguments": arguments}
-        )
-        return self._server.send(packet).body
-
-    def get_status(self) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_STATUS,
-            body={}
-        )
-        return self._server.send(packet).body
-
-    def save_history(self, messages: list) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.SAVE_HISTORY,
-            body={"messages": messages}
-        )
-        response = self._server.send(packet)
-        return response.body.get("data", {})
-
-    def get_history(self) -> list:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_HISTORY,
-            body={}
-        )
-        response = self._server.send(packet)
-        data = response.body.get("data", {})
-        return data.get("history", [])
-    
-    def clear_history(self) -> Dict:
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.SAVE_HISTORY,
-            body={"messages": []}
-        )
-        response = self._server.send(packet)
-        return response.body.get("data", {})
-
-    def get_web_users(self) -> list:
-        """获取 Web 用户列表"""
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_WEB_USERS,
-            body={}
-        )
-        return self._server.send(packet).body.get("users", [])
-
-    def set_web_user(self, username: str, password: str) -> Dict:
-        """设置 Web 用户"""
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.SET_WEB_USER,
-            body={"username": username, "password": password}
-        )
-        return self._server.send(packet).body.get("data", {"success": False})
-
-    def get_full_config(self) -> Dict:
-        """获取完整配置"""
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_CONFIG,
-            body={}
-        )
-        response = self._server.send(packet)
-        return response.body if response.body else {"api_config": {}, "tool_limits": {}}
-
-    def report_web_status(self, running: bool, port: int = 4096) -> Dict:
-        """向后端报告 Web 服务运行状态"""
-        packet = Packet(
-            id=self._next_id(),
-            type=PacketType.GET_WEB_SERVICE_STATUS,
-            body={"running": running, "port": port}
-        )
-        response = self._server.send(packet)
-        return response.body if response.body else {"success": False}
-    
-    def shutdown(self) -> None:
-        self._server.shutdown()
--- a/core/embedded_db.py
+++ b/core/embedded_db.py
@ -1,902 +0,0 @@
-"""
-内嵌图数据库 - 基于SQLite实现
-无需Docker，开箱即用
-"""
-
-import sqlite3
-import hashlib
-import json
-from datetime import datetime
-from pathlib import Path
-from typing import List, Dict, Optional, Any
-
-
-class EmbeddedGraphDB:
-    """内嵌图数据库 - SQLite实现"""
-    
-    def __init__(self, db_path: str = "graph_memory.db"):
-        """
-        初始化数据库
-        
-        Args:
-            db_path: 数据库文件路径
-        """
-        self.db_path = Path(db_path)
-        self.conn = None
-        self._init_db()
-    
-    def _init_db(self):
-        """初始化数据库表"""
-        self.conn = sqlite3.connect(str(self.db_path), check_same_thread=False)
-        self.conn.row_factory = sqlite3.Row
-        
-        cursor = self.conn.cursor()
-        
-        # 创建实体表
-        cursor.execute("""
-            CREATE TABLE IF NOT EXISTS entities (
-                id INTEGER PRIMARY KEY AUTOINCREMENT,
-                name TEXT UNIQUE NOT NULL,
-                type TEXT,
-                mention_count INTEGER DEFAULT 1,
-                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
-                updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
-            )
-        """)
-        
-        # 创建关系表
-        cursor.execute("""
-            CREATE TABLE IF NOT EXISTS relations (
-                id INTEGER PRIMARY KEY AUTOINCREMENT,
-                source_id INTEGER NOT NULL,
-                target_id INTEGER NOT NULL,
-                relation_type TEXT NOT NULL,
-                confidence REAL DEFAULT 1.0,
-                status TEXT DEFAULT 'active',
-                session_id TEXT,
-                turn_id INTEGER,
-                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
-                updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
-                date_bucket TEXT,
-                superseded_by INTEGER,
-                FOREIGN KEY (source_id) REFERENCES entities(id),
-                FOREIGN KEY (target_id) REFERENCES entities(id)
-            )
-        """)
-        
-        # 创建索引
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_entity_name ON entities(name)")
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_entity_type ON entities(type)")
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_relation_source ON relations(source_id)")
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_relation_target ON relations(target_id)")
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_relation_type ON relations(relation_type)")
-        cursor.execute("CREATE INDEX IF NOT EXISTS idx_relation_status ON relations(status)")
-        
-        cursor.execute("""
-            SELECT name FROM sqlite_master 
-            WHERE type='table' AND name='chat_records'
-        """)
-        if not cursor.fetchone():
-            cursor.execute("""
-                CREATE TABLE chat_records (
-                    id INTEGER PRIMARY KEY AUTOINCREMENT,
-                    role TEXT NOT NULL,
-                    content TEXT NOT NULL,
-                    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
-                )
-            """)
-            cursor.execute("CREATE INDEX idx_chat_created ON chat_records(created_at)")
-        
-        # 创建 Web 用户表（支持多用户隔离）
-        cursor.execute("""
-            CREATE TABLE IF NOT EXISTS web_users (
-                id INTEGER PRIMARY KEY AUTOINCREMENT,
-                username TEXT UNIQUE NOT NULL,
-                password_hash TEXT NOT NULL,
-                role TEXT NOT NULL DEFAULT 'user',
-                config_path TEXT,
-                db_path TEXT,
-                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
-                updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
-            )
-        """)
-        
-        # 检查并添加新字段（用于旧数据库迁移）
-        cursor.execute("PRAGMA table_info(web_users)")
-        columns = [row[1] for row in cursor.fetchall()]
-        if 'config_path' not in columns:
-            cursor.execute("ALTER TABLE web_users ADD COLUMN config_path TEXT")
-        if 'db_path' not in columns:
-            cursor.execute("ALTER TABLE web_users ADD COLUMN db_path TEXT")
-        if 'role' not in columns:
-            cursor.execute("ALTER TABLE web_users ADD COLUMN role TEXT NOT NULL DEFAULT 'user'")
-
-        # 确保至少有一个 admin（当 role 列刚添加时，已有用户都是 user）
-        cursor.execute("SELECT COUNT(*) as cnt FROM web_users WHERE role = 'admin'")
-        has_admin = cursor.fetchone()[0] > 0
-        if not has_admin:
-            cursor.execute("SELECT id, username FROM web_users ORDER BY created_at ASC LIMIT 1")
-            first_user = cursor.fetchone()
-            if first_user:
-                cursor.execute("UPDATE web_users SET role = 'admin' WHERE id = ?", (first_user[0],))
-
-        self.conn.commit()
-    
-    def ensure_constraints(self):
-        """确保约束（兼容Neo4j接口）"""
-        pass  # SQLite自动处理
-    
-    def recall(self, query_intent: str, seed_entities: List[str] = None,
-               depth: int = 2, time_range: Dict = None,
-               session_filter: str = None) -> Dict:
-        """
-        检索相关记忆
-        
-        Args:
-            query_intent: 查询关键词（逗号分隔）
-            seed_entities: 种子实体
-            depth: 搜索深度
-            time_range: 时间范围
-            session_filter: 会话过滤
-        
-        Returns:
-            检索结果
-        """
-        keywords = [w.strip().lower() for w in query_intent.replace(',', ' ').split() if w.strip()]
-
-        cursor = self.conn.cursor()
-
-        # 搜索实体
-        entities = []
-        entity_ids = set()
-
-        # 如果没有关键词，返回所有实体（用于"我们都聊过什么"这类问题）
-        if not keywords and not seed_entities:
-            cursor.execute("""
-                SELECT id, name, type, mention_count
-                FROM entities
-                ORDER BY mention_count DESC
-                LIMIT 50
-            """)
-
-            for row in cursor.fetchall():
-                entity_ids.add(row['id'])
-                entities.append({
-                    'name': row['name'],
-                    'type': row['type'] or 'unknown',
-                    'mention_count': row['mention_count']
-                })
-        else:
-            # 有关键词，按关键词搜索
-            for keyword in keywords:
-                cursor.execute("""
-                    SELECT id, name, type, mention_count
-                    FROM entities
-                    WHERE LOWER(name) LIKE ?
-                """, (f"%{keyword}%",))
-
-                for row in cursor.fetchall():
-                    if row['id'] not in entity_ids:
-                        entity_ids.add(row['id'])
-                        entities.append({
-                            'name': row['name'],
-                            'type': row['type'] or 'unknown',
-                            'mention_count': row['mention_count']
-                        })
-        
-        # 广度优先搜索（BFS）扩展实体和关系
-        relations = []
-        visited_entity_ids = set(entity_ids)  # 已访问的实体
-        current_layer_ids = set(entity_ids)   # 当前层的实体
-        
-        # 记录每个实体的深度
-        entity_depths = {}  # entity_id -> depth
-        for eid in entity_ids:
-            entity_depths[eid] = 0
-        
-        for layer in range(depth):
-            if not current_layer_ids:
-                break
-            
-            # 查询当前层实体的所有关系
-            placeholders = ','.join('?' * len(current_layer_ids))
-            
-            query = f"""
-                SELECT r.id, r.source_id, r.target_id,
-                       e1.name as source, e2.name as target,
-                       r.relation_type as type, r.confidence, r.session_id,
-                       r.turn_id, r.created_at, r.status
-                FROM relations r
-                JOIN entities e1 ON r.source_id = e1.id
-                JOIN entities e2 ON r.target_id = e2.id
-                WHERE (r.source_id IN ({placeholders}) OR r.target_id IN ({placeholders}))
-                  AND r.status = 'active'
-            """
-            
-            params = list(current_layer_ids) + list(current_layer_ids)
-            
-            if session_filter:
-                query += " AND r.session_id = ?"
-                params.append(session_filter)
-            
-            cursor.execute(query, params)
-            
-            # 收集下一层的实体
-            next_layer_ids = set()
-            current_layer_relations = []  # 当前层的关系
-            
-            for row in cursor.fetchall():
-                # 计算关系的深度（取两端实体深度的最大值+1）
-                source_depth = entity_depths.get(row['source_id'], layer)
-                target_depth = entity_depths.get(row['target_id'], layer)
-                relation_depth = max(source_depth, target_depth) + 1
-                
-                # 添加关系（带深度标注）
-                current_layer_relations.append({
-                    'source': row['source'],
-                    'target': row['target'],
-                    'type': row['type'],
-                    'confidence': row['confidence'],
-                    'session_id': row['session_id'],
-                    'turn_id': row['turn_id'],
-                    'created_at': row['created_at'],
-                    'status': row['status'],
-                    'depth': relation_depth
-                })
-                
-                # 收集新实体（未访问过的）
-                source_id = row['source_id']
-                target_id = row['target_id']
-                
-                if source_id not in visited_entity_ids:
-                    next_layer_ids.add(source_id)
-                    visited_entity_ids.add(source_id)
-                    entity_depths[source_id] = layer + 1
-                
-                if target_id not in visited_entity_ids:
-                    next_layer_ids.add(target_id)
-                    visited_entity_ids.add(target_id)
-                    entity_depths[target_id] = layer + 1
-            
-            relations.extend(current_layer_relations)
-            
-            # 查询下一层实体的详细信息
-            if next_layer_ids:
-                placeholders = ','.join('?' * len(next_layer_ids))
-                cursor.execute(f"""
-                    SELECT id, name, type, mention_count
-                    FROM entities
-                    WHERE id IN ({placeholders})
-                """, list(next_layer_ids))
-                
-                for row in cursor.fetchall():
-                    entities.append({
-                        'name': row['name'],
-                        'type': row['type'] or 'unknown',
-                        'mention_count': row['mention_count'],
-                        'depth': entity_depths.get(row['id'], layer + 1)
-                    })
-            
-            # 移动到下一层
-            current_layer_ids = next_layer_ids
-        
-        # 为种子实体添加深度标注（depth=0）
-        if entity_ids:
-            # 重新标注种子实体的深度
-            for entity in entities:
-                if entity.get('depth') is None:
-                    entity['depth'] = 0
-        
-        return {
-            "entities": entities,
-            "relations": relations,
-            "message": f"找到 {len(entities)} 个实体, {len(relations)} 条关系"
-        }
-
-    def get_recent_tasks(self, limit: int = 10, state_filter: str = None) -> Dict:
-        """
-        获取最近的任务节点
-
-        Args:
-            limit: 返回数量
-            state_filter: 可选状态过滤（如：进行中、已完成、已暂停、已取消、archived）
-
-        Returns:
-            {"tasks": [{"task_id": str, "description": str, "state": str,
-                         "info_count": int, "updated_at": str}, ...]}
-        """
-        cursor = self.conn.cursor()
-
-        # 查询所有 TaskNode 实体
-        cursor.execute("""
-            SELECT e.id, e.name, e.updated_at
-            FROM entities e
-            WHERE e.type = 'TaskNode'
-            ORDER BY e.updated_at DESC
-            LIMIT ?
-        """, (limit,))
-        rows = cursor.fetchall()
-
-        tasks = []
-        for row in rows:
-            entity_id, name, updated_at = row
-
-            # 查 description
-            cursor.execute("""
-                SELECT r.relation_type, t.name
-                FROM relations r
-                JOIN entities t ON r.target_id = t.id
-                WHERE r.source_id = ? AND r.status = 'active'
-                  AND r.relation_type IN ('has_description', 'HAS_STATE')
-            """, (entity_id,))
-            desc = ""
-            state = "未知"
-            for rtype, tname in cursor.fetchall():
-                if rtype == 'has_description':
-                    desc = tname
-                elif rtype == 'HAS_STATE':
-                    state = tname.replace('State_', '')
-
-            # 可选状态过滤
-            if state_filter and state != state_filter:
-                continue
-
-            # 查关联信息节点数量
-            cursor.execute("""
-                SELECT COUNT(*)
-                FROM relations
-                WHERE source_id = ? AND relation_type = 'CONTAINS_INFO' AND status = 'active'
-            """, (entity_id,))
-            info_count = cursor.fetchone()[0]
-
-            tasks.append({
-                "task_id": name,
-                "description": desc,
-                "state": state,
-                "info_count": info_count,
-                "updated_at": updated_at
-            })
-
-        return {
-            "tasks": tasks,
-            "total": len(tasks)
-        }
-
-    def commit(self, triplets: List[Dict], entity_types: Dict = None,
-               temporal_tag: str = None, session_id: str = None,
-               turn_id: int = None) -> Dict:
-        """
-        写入记忆
-        
-        Args:
-            triplets: 三元组列表
-            entity_types: 实体类型
-            temporal_tag: 时间标签
-            session_id: 会话ID
-            turn_id: 轮次ID
-        
-        Returns:
-            写入结果
-        """
-        cursor = self.conn.cursor()
-        
-        created_entities = 0
-        created_relations = 0
-        
-        for triplet in triplets:
-            subject = triplet.get('subject')
-            relation = triplet.get('relation')
-            obj = triplet.get('object')
-            confidence = triplet.get('confidence', 1.0)
-            
-            if not all([subject, relation, obj]):
-                continue
-            
-            # 创建或更新实体
-            for entity_name, entity_key in [(subject, 'subject_type'), (obj, 'object_type')]:
-                # 按优先级获取实体类型：1) triplet中的_type字段 2) entity_types字典 3) 默认
-                entity_type = triplet.get(entity_key) or (entity_types.get(entity_name) if entity_types else None) or 'Concept'
-                
-                cursor.execute("""
-                    INSERT INTO entities (name, type)
-                    VALUES (?, ?)
-                    ON CONFLICT(name) DO UPDATE SET
-                        mention_count = mention_count + 1,
-                        updated_at = CURRENT_TIMESTAMP
-                """, (entity_name, entity_type))
-                
-                if cursor.rowcount > 0:
-                    created_entities += 1
-            
-            # 获取实体ID
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (subject,))
-            source_id = cursor.fetchone()['id']
-            
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (obj,))
-            target_id = cursor.fetchone()['id']
-            
-            # 创建关系
-            date_bucket = datetime.now().strftime('%Y-%m-%d')
-            
-            cursor.execute("""
-                INSERT INTO relations (
-                    source_id, target_id, relation_type, confidence,
-                    session_id, turn_id, date_bucket
-                )
-                VALUES (?, ?, ?, ?, ?, ?, ?)
-            """, (source_id, target_id, relation, confidence,
-                  session_id, turn_id, date_bucket))
-            
-            created_relations += 1
-        
-        self.conn.commit()
-        
-        return {
-            "created_entities": created_entities,
-            "created_relations": created_relations,
-            "message": f"创建了 {created_entities} 个实体, {created_relations} 条关系"
-        }
-    
-    def purge(self, criteria: Dict, mode: str = "soft",
-              new_relation: Dict = None) -> Dict:
-        """
-        删除或修正记忆
-        
-        Args:
-            criteria: 删除条件
-                支持:
-                - source: 源实体名（精确匹配）
-                - target: 目标实体名（精确匹配）
-                - relation: 关系类型
-                - subject_contains: 源实体名包含（模糊匹配）
-                - target_contains: 目标实体名包含（模糊匹配）
-                - relation_type: 关系类型（同 relation）
-                - source_type: 源实体类型过滤
-                - target_type: 目标实体类型过滤
-                - source_has_status: 源实体 mentions_count 状态（支持 type 字段）
-            mode: 删除模式 (soft/hard)
-            new_relation: 替代关系
-        
-        Returns:
-            删除结果
-        """
-        cursor = self.conn.cursor()
-        
-        # 构建查询条件
-        conditions = []
-        params = []
-        
-        relation_type = criteria.get('relation') or criteria.get('relation_type', '')
-        
-        if criteria.get('source'):
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (criteria['source'],))
-            row = cursor.fetchone()
-            if row:
-                conditions.append("source_id = ?")
-                params.append(row['id'])
-        
-        if criteria.get('target'):
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (criteria['target'],))
-            row = cursor.fetchone()
-            if row:
-                conditions.append("target_id = ?")
-                params.append(row['id'])
-        
-        if relation_type:
-            conditions.append("relation_type = ?")
-            params.append(relation_type)
-        
-        # 通过子查询支持实体属性过滤
-        if criteria.get('subject_contains'):
-            cursor.execute("SELECT id FROM entities WHERE name LIKE ?",
-                          (f'%{criteria["subject_contains"]}%',))
-            ids = [row['id'] for row in cursor.fetchall()]
-            if ids:
-                placeholders = ','.join(['?'] * len(ids))
-                conditions.append(f"source_id IN ({placeholders})")
-                params.extend(ids)
-        
-        if criteria.get('target_contains'):
-            cursor.execute("SELECT id FROM entities WHERE name LIKE ?",
-                          (f'%{criteria["target_contains"]}%',))
-            ids = [row['id'] for row in cursor.fetchall()]
-            if ids:
-                placeholders = ','.join(['?'] * len(ids))
-                conditions.append(f"target_id IN ({placeholders})")
-                params.extend(ids)
-        
-        # 源实体类型过滤
-        if criteria.get('source_type'):
-            cursor.execute("SELECT id FROM entities WHERE type = ?",
-                          (criteria['source_type'],))
-            ids = [row['id'] for row in cursor.fetchall()]
-            if ids:
-                placeholders = ','.join(['?'] * len(ids))
-                conditions.append(f"source_id IN ({placeholders})")
-                params.extend(ids)
-        
-        # 目标实体类型过滤
-        if criteria.get('target_type'):
-            cursor.execute("SELECT id FROM entities WHERE type = ?",
-                          (criteria['target_type'],))
-            ids = [row['id'] for row in cursor.fetchall()]
-            if ids:
-                placeholders = ','.join(['?'] * len(ids))
-                conditions.append(f"target_id IN ({placeholders})")
-                params.extend(ids)
-        
-        # 源实体状态过滤
-        if criteria.get('source_has_status'):
-            status = criteria['source_has_status']
-            cursor.execute("SELECT id FROM entities WHERE type LIKE ?",
-                          (f'%{status}%',))
-            ids = [row['id'] for row in cursor.fetchall()]
-            if ids:
-                placeholders = ','.join(['?'] * len(ids))
-                conditions.append(f"source_id IN ({placeholders})")
-                params.extend(ids)
-        
-        if not conditions:
-            return {"deleted": 0, "message": "无删除条件"}
-        
-        conditions.append("status = 'active'")
-        where_clause = " AND ".join(conditions)
-        
-        if mode == "soft":
-            cursor.execute(f"""
-                UPDATE relations
-                SET status = 'deleted', updated_at = CURRENT_TIMESTAMP
-                WHERE {where_clause}
-            """, params)
-        else:
-            cursor.execute(f"""
-                DELETE FROM relations
-                WHERE {where_clause}
-            """, params)
-        
-        deleted = cursor.rowcount
-        
-        # 删除孤立实体（没有任何关系的数据节点）
-        cursor.execute("""
-            DELETE FROM entities
-            WHERE id NOT IN (
-                SELECT DISTINCT source_id FROM relations
-                UNION
-                SELECT DISTINCT target_id FROM relations
-            )
-        """)
-        deleted_orphans = cursor.rowcount
-        
-        self.conn.commit()
-        
-        return {
-            "deleted": deleted,
-            "deleted_orphans": deleted_orphans,
-            "mode": mode,
-            "message": f"删除了 {deleted} 条关系, {deleted_orphans} 个孤立实体"
-        }
-    
-    def introspect(self, session_id: str = None) -> Dict:
-        """
-        查看会话状态
-        
-        Args:
-            session_id: 会话ID
-        
-        Returns:
-            会话状态
-        """
-        cursor = self.conn.cursor()
-        
-        # 统计实体
-        cursor.execute("SELECT COUNT(*) as count FROM entities")
-        entity_count = cursor.fetchone()['count']
-        
-        # 统计关系
-        cursor.execute("SELECT COUNT(*) as count FROM relations WHERE status = 'active'")
-        relation_count = cursor.fetchone()['count']
-        
-        return {
-            "entity_count": entity_count,
-            "relation_count": relation_count,
-            "session_id": session_id,
-            "message": f"数据库包含 {entity_count} 个实体, {relation_count} 条关系"
-        }
-    
-    def archive(self, days: int = 30) -> Dict:
-        """归档旧关系"""
-        cursor = self.conn.cursor()
-        
-        cursor.execute("""
-            UPDATE relations
-            SET status = 'archived', updated_at = CURRENT_TIMESTAMP
-            WHERE status = 'active'
-              AND created_at < datetime('now', ?)
-        """, (f'-{days} days',))
-        
-        archived = cursor.rowcount
-        self.conn.commit()
-        
-        return {
-            "archived": archived,
-            "message": f"归档了 {archived} 条关系"
-        }
-    
-    def query_archived(self, days: int = None, keyword: str = "") -> Dict:
-        """
-        查询已归档的记忆
-        
-        Args:
-            days: 可选，最近N天内的归档记录
-            keyword: 可选，过滤包含指定关键词的实体名或关系
-        
-        Returns:
-            归档记录列表
-        """
-        cursor = self.conn.cursor()
-        
-        # 基础 SQL：查询已归档的关系及其关联实体
-        conditions = ["r.status = 'archived'"]
-        params = []
-        
-        # 时间范围过滤（最近N天）
-        if days is not None and days > 0:
-            conditions.append("r.updated_at >= datetime('now', ?)")
-            params.append(f'-{days} days')
-        
-        # 关键词过滤（匹配源实体名、目标实体名、关系类型任一）
-        if keyword:
-            # 先找到匹配的实体ID
-            cursor.execute("SELECT id FROM entities WHERE name LIKE ?", (f'%{keyword}%',))
-            matched_ids = [str(row['id']) for row in cursor.fetchall()]
-            
-            if matched_ids:
-                id_list = ','.join(matched_ids)
-                conditions.append(f"(r.source_id IN ({id_list}) OR r.target_id IN ({id_list}) OR r.relation_type LIKE ?)")
-                params.append(f'%{keyword}%')
-            else:
-                conditions.append("r.relation_type LIKE ?")
-                params.append(f'%{keyword}%')
-        
-        where_clause = " AND ".join(conditions)
-        
-        cursor.execute(f"""
-            SELECT r.id, r.relation_type, r.created_at, r.updated_at,
-                   e.name AS source_name, t.name AS target_name
-            FROM relations r
-            JOIN entities e ON r.source_id = e.id
-            JOIN entities t ON r.target_id = t.id
-            WHERE {where_clause}
-            ORDER BY r.updated_at DESC
-            LIMIT 200
-        """, params)
-        
-        rows = cursor.fetchall()
-        
-        results = []
-        for row in rows:
-            results.append({
-                "id": row['id'],
-                "source": row['source_name'],
-                "relation": row['relation_type'],
-                "target": row['target_name'],
-                "archived_at": row['updated_at'],
-                "created_at": row['created_at']
-            })
-        
-        return {
-            "archived_relations": results,
-            "total_relations": len(results),
-            "message": f"找到 {len(results)} 条归档关系"
-        }
-    
-    def cleanup(self, dry_run: bool = True) -> Dict:
-        """清理已删除数据"""
-        cursor = self.conn.cursor()
-        
-        if dry_run:
-            cursor.execute("""
-                SELECT COUNT(*) as count
-                FROM relations
-                WHERE status = 'deleted'
-                  AND updated_at < datetime('now', '-90 days')
-            """)
-            deleted_relations = cursor.fetchone()['count']
-            
-            return {
-                "dry_run": True,
-                "deleted_relations": deleted_relations,
-                "message": f"将删除 {deleted_relations} 条关系"
-            }
-        else:
-            cursor.execute("""
-                DELETE FROM relations
-                WHERE status = 'deleted'
-                  AND updated_at < datetime('now', '-90 days')
-            """)
-            deleted = cursor.rowcount
-            self.conn.commit()
-            
-            return {
-                "dry_run": False,
-                "deleted": deleted,
-                "message": f"删除了 {deleted} 条关系"
-            }
-    
-    def save_chat_records(self, messages: list) -> Dict:
-        """保存聊天记录到数据库"""
-        cursor = self.conn.cursor()
-        
-        saved = 0
-        for msg in messages:
-            role = msg.get("role")
-            content = msg.get("content")
-            if role and content:
-                cursor.execute(
-                    "INSERT INTO chat_records (role, content) VALUES (?, ?)",
-                    (role, content)
-                )
-                saved += 1
-        
-        self.conn.commit()
-        
-        cursor.execute("""
-            DELETE FROM chat_records 
-            WHERE id NOT IN (
-                SELECT id FROM chat_records 
-                ORDER BY id DESC 
-                LIMIT 500
-            )
-        """)
-        self.conn.commit()
-        
-        return {"saved": saved}
-    
-    def get_chat_records(self, limit: int = 500) -> list:
-        """从数据库获取聊天记录"""
-        cursor = self.conn.cursor()
-        cursor.execute("""
-            SELECT role, content FROM chat_records 
-            ORDER BY id ASC LIMIT ?
-        """, (limit,))
-        return [{"role": row[0], "content": row[1]} for row in cursor.fetchall()]
-    
-    def clear_chat_records(self) -> Dict:
-        """清空聊天记录（保留图数据库）"""
-        cursor = self.conn.cursor()
-        cursor.execute("DELETE FROM chat_records")
-        self.conn.commit()
-        return {"cleared": True}
-
-    def set_web_user(self, username: str, password: str, base_dir: str = None, role: str = 'user') -> Dict:
-        """设置或更新 Web 登录用户。password 是明文，自动哈希存储。
-        自动创建用户目录并设置 config_path 和 db_path。
-        role: 'admin' 或 'user'，默认 'user'"""
-        if not username or not password:
-            return {"success": False, "error": "用户名和密码不能为空"}
-        if role not in ('admin', 'user'):
-            return {"success": False, "error": "角色无效 (admin/user)"}
-        
-        import hashlib
-        from pathlib import Path
-        
-        password_hash = hashlib.sha256(password.encode()).hexdigest()
-        
-        # 确定基础目录
-        if base_dir is None:
-            base_dir = Path.home() / ".trulymem"
-        else:
-            base_dir = Path(base_dir)
-        
-        # 创建用户目录
-        user_dir = base_dir / username
-        user_dir.mkdir(parents=True, exist_ok=True)
-        
-        # 设置用户文件路径
-        config_path = str(user_dir / "config.json")
-        db_path = str(user_dir / f"{username}_graph.db")
-        
-        cursor = self.conn.cursor()
-        # 如果是第一个用户，强制设为 admin
-        if self.get_web_users_count() == 0:
-            role = 'admin'
-        cursor.execute("""
-            INSERT INTO web_users (username, password_hash, role, config_path, db_path)
-            VALUES (?, ?, ?, ?, ?)
-            ON CONFLICT(username) DO UPDATE SET
-                password_hash = excluded.password_hash,
-                role = CASE WHEN web_users.role = 'admin' THEN 'admin' ELSE excluded.role END,
-                config_path = COALESCE(web_users.config_path, excluded.config_path),
-                db_path = COALESCE(web_users.db_path, excluded.db_path),
-                updated_at = CURRENT_TIMESTAMP
-        """, (username, password_hash, role, config_path, db_path))
-        self.conn.commit()
-        return {"success": True, "username": username, "role": role, "config_path": config_path, "db_path": db_path}
-
-    def get_web_users(self) -> List[Dict]:
-        """获取所有 Web 用户列表"""
-        cursor = self.conn.cursor()
-        cursor.execute("SELECT id, username, role, config_path, db_path, created_at, updated_at FROM web_users ORDER BY created_at ASC")
-        users = []
-        for row in cursor.fetchall():
-            users.append({
-                "id": row['id'],
-                "username": row['username'],
-                "role": row['role'],
-                "config_path": row['config_path'],
-                "db_path": row['db_path'],
-                "created_at": row['created_at'],
-                "updated_at": row['updated_at']
-            })
-        return users
-    
-    def get_web_user(self, username: str) -> Optional[Dict]:
-        """获取单个 Web 用户信息"""
-        cursor = self.conn.cursor()
-        cursor.execute("""
-            SELECT id, username, role, config_path, db_path, created_at, updated_at
-            FROM web_users WHERE username = ?
-        """, (username,))
-        row = cursor.fetchone()
-        if row:
-            return {
-                "id": row['id'],
-                "username": row['username'],
-                "role": row['role'],
-                "config_path": row['config_path'],
-                "db_path": row['db_path'],
-                "created_at": row['created_at'],
-                "updated_at": row['updated_at']
-            }
-        return None
-    
-    def is_admin(self, username: str) -> bool:
-        """检查用户是否为管理员"""
-        user = self.get_web_user(username)
-        return user is not None and user.get('role') == 'admin'
-    
-    def delete_web_user(self, username: str) -> Dict:
-        """删除 Web 用户（同时保留文件目录）"""
-        if not username:
-            return {"success": False, "error": "用户名不能为空"}
-        cursor = self.conn.cursor()
-        cursor.execute("DELETE FROM web_users WHERE username = ?", (username,))
-        self.conn.commit()
-        if cursor.rowcount > 0:
-            return {"success": True, "username": username}
-        return {"success": False, "error": "用户不存在"}
-
-    def get_web_users_count(self) -> int:
-        """获取 Web 用户数量 (用于判断是否需要首次设置)"""
-        cursor = self.conn.cursor()
-        cursor.execute("SELECT COUNT(*) as cnt FROM web_users")
-        row = cursor.fetchone()
-        return row['cnt'] if row else 0
-
-    def verify_web_user(self, username: str, password: str) -> bool:
-        """验证 Web 用户登录"""
-        import hashlib
-        password_hash = hashlib.sha256(password.encode()).hexdigest()
-        cursor = self.conn.cursor()
-        cursor.execute("""
-            SELECT id FROM web_users
-            WHERE username = ? AND password_hash = ?
-        """, (username, password_hash))
-        return cursor.fetchone() is not None
-    
-    def close(self):
-        """关闭数据库连接"""
-        if self.conn:
-            self.conn.close()
-            self.conn = None
-    
-    def __enter__(self):
-        return self
-    
-    def __exit__(self, exc_type, exc_val, exc_tb):
-        self.close()
-
-
-# 兼容性别名
-Neo4jGraph = EmbeddedGraphDB
-
--- a/core/graph_client.py
+++ b/core/graph_client.py
@ -1,461 +0,0 @@
-#!/usr/bin/env python3
-"""
-Graph Memory Client - 图记忆客户端核心实现（重构版）
-使用模块化的工具和提示词系统
-"""
-
-import json
-import os
-import uuid
-from datetime import datetime
-from openai import OpenAI
-
-from .tools import TOOLS
-from .tool_executor import execute_tool
-from .prompts.prompt_manager import PromptManager
-
-# 环境配置
-DEEPSEEK_API_KEY = os.environ.get("DEEPSEEK_API_KEY", "")
-DEEPSEEK_BASE_URL = os.environ.get("DEEPSEEK_BASE_URL", "https://api.deepseek.com")
-MODEL_NAME = os.environ.get("MODEL_NAME", "deepseek-v4-flash")
-
-NEO4J_URI = os.environ.get("NEO4J_URI", "bolt://localhost:7687")
-NEO4J_USER = os.environ.get("NEO4J_USER", "neo4j")
-NEO4J_PASSWORD = os.environ.get("NEO4J_PASSWORD", "neo4j")
-
-# 会话配置
-CURRENT_SESSION_ID = f"session-{datetime.now().strftime('%Y%m%d')}-{uuid.uuid4().hex[:4]}"
-CURRENT_TURN = 0
-
-
-class Neo4jGraph:
-    """Neo4j图数据库客户端"""
-    
-    def __init__(self, uri: str, user: str, password: str):
-        from neo4j import GraphDatabase
-        self.driver = GraphDatabase.driver(uri, auth=(user, password))
-    
-    def close(self):
-        self.driver.close()
-    
-    def ensure_constraints(self):
-        """确保约束和索引存在"""
-        with self.driver.session() as session:
-            # 实体约束
-            session.run("CREATE CONSTRAINT entity_name_constraint IF NOT EXISTS FOR (e:Entity) REQUIRE e.name IS UNIQUE")
-            session.run("CREATE CONSTRAINT session_id_constraint IF NOT EXISTS FOR (s:Session) REQUIRE s.session_id IS UNIQUE")
-            
-            # 关系索引
-            session.run("CREATE INDEX rel_created_at IF NOT EXISTS FOR ()-[r:RELATES]-() ON r.created_at")
-            session.run("CREATE INDEX rel_session_id IF NOT EXISTS FOR ()-[r:RELATES]-() ON r.session_id")
-            session.run("CREATE INDEX rel_type IF NOT EXISTS FOR ()-[r:RELATES]-() ON r.type")
-            session.run("CREATE INDEX rel_status IF NOT EXISTS FOR ()-[r:RELATES]-() ON r.status")
-            session.run("CREATE INDEX rel_date_bucket IF NOT EXISTS FOR ()-[r:RELATES]-() ON r.date_bucket")
-            
-            # 实体索引
-            session.run("CREATE INDEX entity_type IF NOT EXISTS FOR (e:Entity) ON e.type")
-            session.run("CREATE INDEX entity_mention_count IF NOT EXISTS FOR (e:Entity) ON e.mention_count")
-    
-    def recall(self, query_intent: str, seed_entities: list = None, depth: int = 2, 
-               time_range: dict = None, session_filter: str = None) -> dict:
-        """检索记忆"""
-        with self.driver.session() as session:
-            # 支持逗号分隔的多个关键词
-            keywords = [w.strip() for w in query_intent.replace(',', ' ').split() if len(w.strip()) > 0]
-            
-            if not keywords and not seed_entities:
-                return {"entities": [], "relations": [], "message": "无查询关键词"}
-            
-            params = {}
-            cond_parts = ["r.status = 'active'"]
-            
-            if session_filter:
-                cond_parts.append("r.session_id = $session_id")
-                params["session_id"] = session_filter
-            
-            if keywords:
-                keyword_conditions = []
-                for k in keywords:
-                    k_lower = k.lower()
-                    keyword_conditions.append(f"toLower(e.name) CONTAINS '{k_lower}'")
-                    keyword_conditions.append(f"toLower(t.name) CONTAINS '{k_lower}'")
-                    keyword_conditions.append(f"toLower(r.type) CONTAINS '{k_lower}'")
-                cond_parts.append(f"({' OR '.join(keyword_conditions)})")
-            
-            if seed_entities:
-                placeholders = ",".join([f"'{s}'" for s in seed_entities])
-                cond_parts.append(f"(e.name IN [{placeholders}] OR t.name IN [{placeholders}])")
-            
-            if time_range and "days" in time_range:
-                cond_parts.append(f"r.created_at >= datetime() - duration('P{time_range['days']}D')")
-            
-            where_clause = " AND ".join(cond_parts)
-            
-            cypher = f"""
-            MATCH (e:Entity)-[r:RELATES]->(t:Entity)
-            WHERE {where_clause}
-            RETURN e, r, t
-            ORDER BY r.created_at DESC
-            LIMIT 30
-            """
-            
-            result = session.run(cypher, params)
-            entities, relations = {}, []
-            
-            for record in result:
-                e, r, t = record["e"], record["r"], record["t"]
-                if e["name"] not in entities:
-                    entities[e["name"]] = {"name": e["name"], "type": e.get("type", "unknown"), "mention_count": e.get("mention_count", 1)}
-                if t["name"] not in entities:
-                    entities[t["name"]] = {"name": t["name"], "type": t.get("type", "unknown"), "mention_count": t.get("mention_count", 1)}
-                
-                relations.append({
-                    "source": e["name"],
-                    "target": t["name"],
-                    "type": r["type"],
-                    "created_at": str(r.get("created_at", "")),
-                    "session_id": r.get("session_id", ""),
-                    "turn_id": r.get("turn_id", 0),
-                    "confidence": r.get("confidence", 1.0)
-                })
-            
-            return {"entities": list(entities.values()), "relations": relations[:20]}
-    
-    def commit(self, triplets: list, entity_types: dict = None, temporal_tag: str = None) -> dict:
-        """写入记忆"""
-        global CURRENT_TURN
-        with self.driver.session() as session:
-            valid_triplets = [t for t in triplets if t.get("subject") and t.get("relation") and t.get("object")]
-            
-            if not valid_triplets:
-                return {"committed_count": 0, "details": []}
-            
-            date_bucket = temporal_tag or datetime.now().strftime("%Y-%m-%d")
-            
-            results = []
-            for triplet in valid_triplets:
-                subject = triplet.get("subject", "").strip()
-                relation = triplet.get("relation", "").strip()
-                obj = triplet.get("object", "").strip()
-                confidence = triplet.get("confidence", 0.9)
-                
-                # 按优先级获取实体类型：1) triplet中的_type字段 2) entity_types字典 3) 默认
-                s_type = triplet.get("subject_type") or (entity_types.get(subject) if entity_types else None) or "Concept"
-                o_type = triplet.get("object_type") or (entity_types.get(obj) if entity_types else None) or "Concept"
-                
-                session.run("""
-                MERGE (s:Entity {name: $subject})
-                ON CREATE SET s.type = $s_type, s.created_at = datetime(), s.mention_count = 1, s.updated_at = datetime()
-                ON MATCH SET s.mention_count = coalesce(s.mention_count, 0) + 1, s.updated_at = datetime()
-                
-                MERGE (t:Entity {name: $object})
-                ON CREATE SET t.type = $o_type, t.created_at = datetime(), t.mention_count = 1, t.updated_at = datetime()
-                ON MATCH SET t.mention_count = coalesce(t.mention_count, 0) + 1, t.updated_at = datetime()
-                
-                CREATE (s)-[r:RELATES {
-                    type: $relation,
-                    created_at: datetime(),
-                    session_id: $session_id,
-                    turn_id: $turn_id,
-                    role: 'user',
-                    status: 'active',
-                    confidence: $confidence,
-                    date_bucket: $date_bucket
-                }]->(t)
-                """, subject=subject, object=obj, relation=relation,
-                      s_type=s_type, o_type=o_type,
-                      session_id=CURRENT_SESSION_ID, turn_id=CURRENT_TURN, confidence=confidence,
-                      date_bucket=date_bucket)
-                
-                results.append(f"{subject} -[{relation}]-> {obj}")
-            
-            return {"committed_count": len(results), "details": results}
-    
-    def purge(self, criteria: dict, mode: str = "soft", new_relation: dict = None) -> dict:
-        """删除记忆"""
-        with self.driver.session() as session:
-            subject_pattern = criteria.get("subject_contains", "")
-            rel_type = criteria.get("relation_type", "")
-            target_pattern = criteria.get("target_contains", "")
-            source_type = criteria.get("source_type", "")
-            target_type = criteria.get("target_type", "")
-            source_status = criteria.get("source_has_status", "")
-            session_id = criteria.get("session_id", CURRENT_SESSION_ID)
-            
-            cond_parts = ["r.status = 'active'"]
-            params = {"session_id": session_id}
-            
-            if subject_pattern:
-                cond_parts.append("e.name CONTAINS $subject")
-                params["subject"] = subject_pattern
-            if target_pattern:
-                cond_parts.append("t.name CONTAINS $target")
-                params["target"] = target_pattern
-            if rel_type:
-                cond_parts.append("r.type = $rel_type")
-                params["rel_type"] = rel_type
-            if source_type:
-                cond_parts.append("s.entity_type = $source_type")
-                params["source_type"] = source_type
-            if target_type:
-                cond_parts.append("t.entity_type = $target_type")
-                params["target_type"] = target_type
-            if source_status:
-                cond_parts.append("s.status = $source_status")
-                params["source_status"] = source_status
-            
-            where_clause = " AND ".join(cond_parts)
-            
-            if mode == "supersede" and new_relation:
-                new_rel = new_relation.get("relation", "")
-                new_target = new_relation.get("target", "")
-                
-                if not new_rel or not new_target:
-                    return {"error": "supersede模式需要提供new_relation.relation和new_relation.target"}
-                
-                result = session.run(f"""
-                MATCH (s:Entity)-[r:RELATES]->(t:Entity)
-                WHERE {where_clause}
-                SET r.status = 'superseded', r.updated_at = datetime()
-                RETURN count(r) as count
-                """, params)
-                
-                count = result.single()["count"]
-                return {"deleted_count": count, "mode": "supersede"}
-            else:
-                result = session.run(f"""
-                MATCH (s:Entity)-[r:RELATES]->(t:Entity)
-                WHERE {where_clause}
-                SET r.status = 'deleted', r.updated_at = datetime()
-                RETURN count(r) as deleted
-                """, params)
-                count = result.single()["deleted"]
-                
-                # 删除孤立节点（没有任何关系的实体）
-                orphan_result = session.run("""
-                MATCH (e:Entity)
-                WHERE NOT (e)-[:RELATES]-()
-                DELETE e
-                RETURN count(e) as orphans
-                """)
-                orphan_count = orphan_result.single()["orphans"]
-                
-                return {"deleted_count": count, "orphan_count": orphan_count, "mode": "soft"}
-    
-    def introspect(self, session_id: str = None) -> dict:
-        """查看记忆状态"""
-        target_session = session_id or CURRENT_SESSION_ID
-        
-        with self.driver.session() as session:
-            result = session.run("""
-            MATCH (s:Entity)-[r:RELATES]->(t:Entity)
-            WHERE r.session_id = $session_id AND r.status = 'active'
-            RETURN collect(DISTINCT s.name) as source_entities,
-                   collect(DISTINCT t.name) as target_entities,
-                   count(r) as rel_count,
-                   collect(DISTINCT r.type) as rel_types
-            """, session_id=target_session)
-            record = result.single()
-            
-            result2 = session.run("""
-            MATCH (e:Entity)
-            RETURN e.name as name, e.mention_count as count, e.type as type
-            ORDER BY e.mention_count DESC
-            LIMIT 10
-            """)
-            hotspots = [(r["name"], r["count"], r["type"]) for r in result2]
-            
-            return {
-                "session_id": target_session,
-                "total_turns": CURRENT_TURN,
-                "entities_discussed": list(set((record["source_entities"] or []) + (record["target_entities"] or []))),
-                "relation_count": record["rel_count"] if record else 0,
-                "relation_types": record["rel_types"] if record else [],
-                "memory_hotspots": hotspots
-            }
-    
-    def archive(self, days: int = 30) -> dict:
-        """归档旧记忆"""
-        with self.driver.session() as session:
-            result = session.run("""
-            MATCH ()-[r:RELATES]->()
-            WHERE r.status = 'active' AND r.created_at < datetime() - duration('P' + $days + 'D')
-            SET r.status = 'archived', r.archived_at = datetime()
-            RETURN count(r) as archived
-            """, days=str(days))
-            
-            return {"archived_count": result.single()["archived"], "days": days}
-    
-    def query_archived(self, days: int = None, keyword: str = "") -> dict:
-        """查询归档记忆"""
-        with self.driver.session() as session:
-            filters = []
-            params = {}
-            
-            filters.append("r.status = 'archived'")
-            
-            if days is not None and days > 0:
-                filters.append("r.archived_at >= datetime() - duration('P' + $days + 'D')")
-                params["days"] = str(days)
-            
-            if keyword:
-                filters.append("(e.name CONTAINS $keyword OR t.name CONTAINS $keyword OR r.type CONTAINS $keyword)")
-                params["keyword"] = keyword
-            
-            where = " AND ".join(filters)
-            
-            result = session.run(f"""
-            MATCH (e:Entity)-[r:RELATES]->(t:Entity)
-            WHERE {where}
-            RETURN e.name as source, r.type as relation, t.name as target,
-                   r.archived_at as archived_at, r.created_at as created_at
-            ORDER BY r.archived_at DESC
-            LIMIT 200
-            """, params)
-            
-            records = []
-            for row in result:
-                records.append({
-                    "source": row["source"],
-                    "relation": row["relation"],
-                    "target": row["target"],
-                    "archived_at": str(row["archived_at"]) if row.get("archived_at") else "",
-                    "created_at": str(row["created_at"]) if row.get("created_at") else ""
-                })
-            
-            return {
-                "archived_relations": records,
-                "total_relations": len(records),
-                "message": f"找到 {len(records)} 条归档关系"
-            }
-    
-    def cleanup(self, dry_run: bool = True) -> dict:
-        """清理无效数据"""
-        with self.driver.session() as session:
-            result1 = session.run("""
-            MATCH ()-[r:RELATES]->()
-            WHERE r.status = 'deleted' AND r.updated_at < datetime() - duration('P90D')
-            RETURN count(r) as to_delete
-            """)
-            deleted_relations = result1.single()["to_delete"]
-            
-            result2 = session.run("""
-            MATCH (e:Entity)
-            WHERE NOT (e)-[:RELATES]-()
-            RETURN count(e) as orphans
-            """)
-            orphan_nodes = result2.single()["orphans"]
-            
-            if not dry_run and deleted_relations > 0:
-                session.run("""
-                MATCH ()-[r:RELATES]->()
-                WHERE r.status = 'deleted' AND r.updated_at < datetime() - duration('P90D')
-                DELETE r
-                """)
-            
-            if not dry_run and orphan_nodes > 0:
-                session.run("""
-                MATCH (e:Entity)
-                WHERE NOT (e)-[:RELATES]-()
-                DELETE e
-                """)
-            
-            return {
-                "dry_run": dry_run,
-                "deleted_relations": deleted_relations,
-                "orphan_nodes": orphan_nodes,
-                "action_taken": not dry_run
-            }
-
-
-class GraphMemoryClient:
-    """图记忆客户端"""
-    
-    def __init__(self, api_key: str, base_url: str, graph, model: str = "deepseek-v4-flash"):
-        # 清理可能存在的错误代理环境变量
-        import os
-        proxy_vars = ['http_proxy', 'https_proxy', 'HTTP_PROXY', 'HTTPS_PROXY', 'all_proxy', 'ALL_PROXY']
-        for var in proxy_vars:
-            if var in os.environ:
-                value = os.environ[var]
-                # 如果代理URL没有scheme前缀，添加http://
-                if value and not value.startswith(('http://', 'https://', 'socks5://', 'socks4://')):
-                    os.environ[var] = f'http://{value}'
-        
-        self.client = OpenAI(api_key=api_key, base_url=base_url)
-        self.graph = graph
-        self.tools = TOOLS
-        self.model = model
-        
-        prompt_manager = PromptManager()
-        self.system_prompt = prompt_manager.get_system_prompt()
-    
-    def send_message(self, user_input: str, tool_results: list = None, assistant_msg: dict = None) -> dict:
-        """发送消息"""
-        global CURRENT_TURN
-        
-        messages = [{"role": "system", "content": self.system_prompt}]
-        
-        # 添加用户消息
-        messages.append({"role": "user", "content": user_input})
-        
-        # 添加 assistant 消息（包含 tool_calls）
-        if assistant_msg:
-            messages.append(assistant_msg)
-        
-        # 添加工具结果
-        if tool_results:
-            messages.extend(tool_results)
-        
-        response = self.client.chat.completions.create(
-            model=self.model,
-            messages=messages,
-            tools=self.tools,
-            tool_choice="auto"
-        )
-        
-        return response
-    
-    def send_message_with_history(self, messages_history: list) -> dict:
-        """使用消息历史发送消息"""
-        global CURRENT_TURN
-        
-        # 构建完整消息列表
-        messages = [{"role": "system", "content": self.system_prompt}]
-        messages.extend(messages_history)
-        
-        response = self.client.chat.completions.create(
-            model=self.model,
-            messages=messages,
-            tools=self.tools,
-            tool_choice="auto"
-        )
-        
-        return response
-
-    def send_message_stream(self, user_input: str, tool_results: list = None, assistant_msg: dict = None):
-        """流式发送消息"""
-        global CURRENT_TURN
-
-        messages = [{"role": "system", "content": self.system_prompt}]
-
-        # 添加用户消息
-        messages.append({"role": "user", "content": user_input})
-
-        # 添加 assistant 消息（包含 tool_calls）
-        if assistant_msg:
-            messages.append(assistant_msg)
-
-        # 添加工具结果
-        if tool_results:
-            messages.extend(tool_results)
-
-        stream = self.client.chat.completions.create(
-            model=self.model,
-            messages=messages,
-            tools=self.tools,
-            tool_choice="auto",
-            stream=True
-        )
-
-        return stream
--- a/core/migrate.py
+++ b/core/migrate.py
@ -1,135 +0,0 @@
-"""
-自动迁移模块 - 从旧版单用户架构迁移到多用户隔离架构
-"""
-
-import os
-import shutil
-import json
-import hashlib
-from pathlib import Path
-from typing import Dict, Optional
-from datetime import datetime
-
-
-def _trulymem_dir() -> Path:
-    return Path.home() / ".trulymem"
-
-def _old_config_path() -> Path:
-    return _trulymem_dir() / "config.json"
-
-def _old_db_path() -> Path:
-    return _trulymem_dir() / "graph_memory.db"
-
-def _new_global_db_path() -> Path:
-    return _trulymem_dir() / "trulymem.db"
-
-def _migrated_flag() -> Path:
-    return _trulymem_dir() / ".migrated"
-
-
-def need_migration() -> bool:
-    """检测是否需要迁移"""
-    # 如果已经迁移过，不需要再迁移
-    if is_migrated():
-        return False
-    
-    old_config_exists = _old_config_path().exists()
-    old_db_exists = _old_db_path().exists()
-    new_db_exists = _new_global_db_path().exists()
-    if (old_config_exists or old_db_exists) and not new_db_exists:
-        return True
-    
-    return False
-
-
-def is_migrated() -> bool:
-    """检查是否已完成迁移"""
-    return _migrated_flag().exists()
-
-
-def _mark_migrated():
-    """标记迁移完成"""
-    _trulymem_dir().mkdir(parents=True, exist_ok=True)
-    with open(_migrated_flag(), 'w') as f:
-        f.write(datetime.now().isoformat())
-
-
-def run_migration(username: str, password: str) -> Dict:
-    """
-    执行迁移
-    
-    Args:
-        username: 新用户名
-        password: 新用户密码
-    
-    Returns:
-        迁移结果字典
-    """
-    try:
-        # 1. 创建用户目录
-        user_dir = _trulymem_dir() / username
-        user_dir.mkdir(parents=True, exist_ok=True)
-        new_config_path = user_dir / "config.json"
-        if _old_config_path().exists():
-            shutil.copy2(_old_config_path(), new_config_path)
-        new_db_path = user_dir / f"{username}_graph.db"
-        if _old_db_path().exists():
-            shutil.copy2(_old_db_path(), new_db_path)
-        
-        # 4. 创建全局数据库并写入 web_users 表
-        from .embedded_db import EmbeddedGraphDB
-        
-        global_db = EmbeddedGraphDB(db_path=str(_new_global_db_path()))
-        
-        # 设置用户（会自动创建记录）
-        result = global_db.set_web_user(username, password)
-        if not result.get("success"):
-            return {"success": False, "error": f"创建用户失败: {result.get('error')}"}
-        
-        # 如果用户目录已存在，更新路径（确保正确）
-        cursor = global_db.conn.cursor()
-        config_path = str(new_config_path)
-        db_path = str(new_db_path)
-        cursor.execute("""
-            UPDATE web_users 
-            SET config_path = ?, db_path = ?
-            WHERE username = ?
-        """, (config_path, db_path, username))
-        global_db.conn.commit()
-        
-        # 5. 标记迁移完成
-        _mark_migrated()
-        
-        global_db.close()
-        
-        return {
-            "success": True,
-            "username": username,
-            "config_path": config_path,
-            "db_path": db_path,
-            "message": "迁移完成"
-        }
-    
-    except Exception as e:
-        return {"success": False, "error": str(e)}
-
-
-def rollback_migration():
-    """回滚迁移（用于失败恢复）"""
-    try:
-        # 删除全局数据库
-        if _new_global_db_path().exists():
-            _new_global_db_path().unlink()
-        if _migrated_flag().exists():
-            _migrated_flag().unlink()
-        
-        return {"success": True, "message": "回滚完成"}
-    except Exception as e:
-        return {"success": False, "error": str(e)}
-
-
-if __name__ == '__main__':
-    # 测试
-    print("Migration module test")
-    print(f"Need migration: {need_migration()}")
-    print(f"Is migrated: {is_migrated()}")
--- a/core/prompts/init.py
+++ b/core/prompts/init.py
@ -1,6 +0,0 @@
-"""
-提示词管理模块
-"""
-from .prompt_manager import PromptManager
-
-__all__ = ["PromptManager"]
--- a/core/prompts/prompt_manager.py
+++ b/core/prompts/prompt_manager.py
@ -1,77 +0,0 @@
-"""
-提示词管理器
-"""
-from pathlib import Path
-
-
-class PromptManager:
-    """提示词管理器"""
-    
-    _instance = None
-    _cached_prompt = None
-    
-    def __new__(cls):
-        if cls._instance is None:
-            cls._instance = super().__new__(cls)
-        return cls._instance
-    
-    def __init__(self):
-        if not hasattr(self, '_initialized'):
-            self.prompts_dir = Path(__file__).parent / "templates"
-            self._initialized = True
-    
-    def get_system_prompt(self) -> str:
-        if PromptManager._cached_prompt is not None:
-            return PromptManager._cached_prompt
-        
-        prompt_file = self.prompts_dir / "system_prompt.md"
-        if prompt_file.exists():
-            with open(prompt_file, "r", encoding="utf-8") as f:
-                PromptManager._cached_prompt = f.read()
-        else:
-            PromptManager._cached_prompt = self._build_default_prompt()
-        
-        return PromptManager._cached_prompt
-    
-    def _build_default_prompt(self) -> str:
-        return """你是TrulyMEM，一个拥有长期记忆能力的AI助手。
-
-## 核心能力
-
-1. **长期记忆** - 基于图数据库存储实体关系
-2. **人设管理** - 支持角色扮演和性格设定
-3. **任务跟踪** - 维护工作记忆链，跟踪连续性任务
-
-## 记忆原则
-
- **明确内容必须写入** - 用户明确提到的信息必须存储
- **推理内容必须标注** - AI推理得到的内容标注[猜测]
- **图数据库是唯一记忆源** - 没有其他记忆方式
-
-## 工具使用
-
-### 记忆工具
- `memory_recall` - 检索记忆
- `memory_commit` - 写入记忆
- `memory_purge` - 删除记忆
- `memory_introspect` - 查看状态
-
-### 人设工具
- `persona_update` - 更新人设
- `persona_clear` - 清除人设
-
-### 任务工具
- `task_create` - 创建任务
- `task_set_state` - 设置状态
- `task_delete` - 删除任务
- `task_link_info` - 关联信息
-
-## 自主性
-
-你有权根据对话上下文自主决定：
- 是否需要查询记忆
- 是否需要写入记忆
- 是否需要维护任务链
- 如何使用工具
-
-记住：灵活应对，保持自然对话体验。"""
--- a/core/prompts/templates/system_prompt.md
+++ b/core/prompts/templates/system_prompt.md
@ -1,183 +0,0 @@
-# TrulyMEM 系统提示词
-
-你是TrulyMEM，一个拥有长期记忆能力的AI助手。
-
-**人设兜底规则**：当图数据库中没有查到人设信息时，以「我是 TrulyMEM，一个有长期记忆的 AI 助手」作为默认开场。如果人设图返回了角色信息，按人设执行即可。
-
-## ⚠️ 强制执行顺序（内部流程，不得向用户输出）
-
-**以下步骤是内部流程，绝对不要在你的回复中提及或输出。** 你应当仅通过工具调用悄悄完成，回复时直接给出自然的对话内容。
-
-步骤1：memory_recall 查询人设图和相关长期记忆
-步骤2：task_query 查询工作记忆链和最近任务
-步骤3：结合上下文处理用户输入并形成回复思路
-步骤4：memory_commit (写入关键信息)
-步骤5：task_archive 归档已完成或过期任务；若本轮查询类工具调用 ≥5 次，再单独调用 context_rewrite 压缩工具 JSON
-
-**违反规则的后果**：
- 输出步骤内容 → 暴露内部机制，用户体验极差，违反最高优先级指令
- 跳过步骤1 → 无法获取人设
- 跳过步骤5 → 工作记忆无限膨胀
-
---
-
-## ⚠️ 最高优先级：工具执行期间禁止输出
-
-**在完成所有工具调用之前，绝对不要输出任何文字。**
-
-正确流程：
-1. 调用所有必要的工具（memory_recall、task_query 等）→ **不输出任何文字**
-2. 等所有工具返回结果 → **仍然不输出任何文字**
-3. 处理返回结果，思考回复内容 → **仍然不输出任何文字**
-4. **最后，只输出一次完整的回复**
-
-**禁止的行为**：
- ❌ 先输出「你好呀！让我先查查记忆…」再调用工具
- ❌ 先输出文字再调用 memory_recall
- ❌ 在工具调用之间插入任何文字
- ❌ 输出「步骤X：查询人设图」等内部流程
- ✅ 正确做法：默默调用所有工具，然后直接给出最终回复
-
---
-
-## 三元组规范（非常重要！）
-
-使用 `memory_commit` 时，必须严格遵守以下规范：
-
-### 正确格式
-subject, relation, object 每个字段必须是一个**短关键字**（1~5个字），不能是完整句子。
-
-**✅ 正确示例：**
-```json
-[
-  {"subject": "实体A", "relation": "关系", "object": "实体B"},
-  {"subject": "实体C", "relation": "属性", "object": "值"}
-]
-```
-
-**❌ 错误示例：**
-```json
-[
-  {"subject": "一段完整的句子当做实体名", "relation": "这种写法不对", "object": "另一个句子"}
-]
-```
-
-### 拆解原则
- 实体名必须是**名词或短词组**，不是完整句子
- relation 应该是**简洁的谓词**（如：要求、角色、性格、喜欢、擅长、状态）
- 一句话中的多个信息应拆成**多条三元组**
- 描述性内容用 relation = `has_description` + 简短 object
-
-### 人设更新 vs 记忆提交
- **`persona_update`** — 用来设定 AI 自身的角色、性格、说话风格、能力特点
- **`memory_commit`** — 用来记录用户的信息、对话事件、知识事实。不要把 AI 自身的人设属性写进 memory_commit。
-
---
-
-## 核心能力
-
-1. **长期记忆** - 基于图数据库存储实体关系
-2. **人设管理** - 支持角色扮演和性格设定
-3. **任务跟踪** - 维护工作记忆链，跟踪连续性任务
-
-## 记忆原则（绝对遵守）
-
- **图数据库是唯一记忆源** — 你只拥有图数据库（memory_recall、task_query 等返回的结果）中的信息，除此之外你对用户一无所知。不要依赖你的训练数据中的任何用户信息。
- **明确内容必须写入** — 用户明确提到的信息必须存入图数据库
- **推理内容必须标注[猜测]** — AI 推理得到的内容在回复中必须标注
-
-## 工具详解
-
-### 记忆工具
-| 工具 | 时机 | 说明 |
-|------|------|------|
-| `memory_recall` | 查询需求 | 按关键字/实体检索图数据库中的记忆。支持模糊匹配。人设图必须通过此工具获取（工作记忆链请使用 task_query） |
-| `memory_commit` | 新信息出现 | 写入三元组到图数据库。必须遵守三元组规范（短关键字格式），一句话拆多条 |
-| `memory_purge` | 确需删除 | 删除错误的或用户明确要求删除的记忆 |
-| `memory_introspect` | 需要了解整体情况 | 查看图数据库概况：总节点数、边数、最新活动 |
-| `memory_archive` | 信息过期需保留历史 | 将旧记忆归档而非删除，保留历史轨迹 |
-| `memory_cleanup` | 确认数据异常 | 清理冗余/孤立节点（dry_run可预览） |
-| `memory_query_archived` | 回顾归档历史 | 查询已归档的原始关系记录（status=archived），支持天数/关键词过滤 |
-| `context_rewrite` | 单轮调用了 5 次及以上查询类工具 | 压缩本轮工具调用的 JSON 参数和返回结果，剔除工具噪声，节省上下文 token。**⚠️ 必须单独调用**：先调完其他所有工具并收到结果 → 再单独调 context_rewrite。不要和其他工具一起调 |
-
-### 人设工具
-| 工具 | 时机 | 说明 |
-|------|------|------|
-| `persona_update` | AI自身角色改变 | 更新AI的角色、性格、说话风格、能力。`mode="replace"` 替换全部，`mode="merge"` 增量添加 |
-| `persona_remove` | 只需删除某一条属性 | 删除单条人设属性（如只删除说话风格，保留扮演角色不变） |
-| `persona_clear` | 需要完全重置 | 清除所有AI人设属性。**此操作不可逆，需要 confirm=true** |
-
-### 任务工具（生命周期管理）
-| 工具 | 时机 | 说明 |
-|------|------|------|
-| `task_query` | 新对话/需要回顾 | 查询最近任务列表（按更新时间倒序）。**新对话开始时优先调用此工具**，了解现有任务后再决定是继续还是创建新任务 |
-| `task_create` | 用户提出实质性话题后 | 创建任务节点。**不要在纯问候/打招呼时创建任务**——等用户说出具体话题后再创建。判断标准：用户消息是否包含可讨论的具体内容 |
-| `task_set_state` | 状态变更 | 修改任务状态（active、completed、archived）。**旧会话结束后必须将对应的任务设为 archived** |
-| `task_archive` | 强制执行顺序的步骤5 | 归档已完成/过期的任务。将任务状态设为 archived，同时写入完成摘要到图数据库。**每轮对话最后必须检查是否需要调用此工具** |
-| `task_delete` | 确需删除的任务 | 彻底删除任务节点 |
-| `task_link_info` | 信息归属 | 将记忆节点关联到特定的任务。**只关联到相关的任务，不要全部链到「当前轮对话」** |
-
---
-
-## ⚠️ 任务生命周期规范（避免记忆膨胀）
-
-AI 最常见的错误是：**把每一轮的所有节点都关联到「当前轮对话」，但从不归档过时的任务，导致图数据库无限膨胀。**
-
-### 正确做法
-
-```
-1. 新会话开始 → task_create 创建「当前轮对话-<时间/主题>」
-2. 对话过程中 → 根据实际归属使用 task_link_info
-3. 话题结束/转变时 → task_archive 归档旧任务（代替 task_set_state）
-4. 归档后 → 再创建新的当前轮对话任务
-```
-
-### 步骤5 归档规则（强制）
-
-每轮对话最后一步 check 现有任务：
- **已完成的任务** → 调用 `task_archive` 归档，写入完成摘要
- **长时间无更新的任务**（>3轮对话） → 调用 `task_archive` 归档
- **topic 已转变** → 旧任务归档，新任务创建
- **所有 active 任务超过3个** → 归档最旧的
-
-> 即使本轮没有主题转变，也应按需检查归档状态。
-> `task_archive` 比 `task_set_state(state=archived)` 多一个写入完成摘要的功能，优先使用。
-
-### 绝对禁止
- ❌ 把每一条记忆都链到同一个「当前轮对话」任务
- ❌ 跳过步骤5（从不归档）导致工作记忆无限膨胀
- ❌ 对同一个任务堆积数千条关联
- ❌ 使用 task_delete 代替归档（归档保留历史，删除丢失上下文）
-
-### 生命流程示例
-
-**第1轮：**
-```
-task_create(描述="当前轮对话-工作规划", state=active)
-memory_commit(用户说春节计划)
-task_link_info(info="春节计划", task="当前轮对话-工作规划")
-task_archive(task="当前轮对话-工作规划", summary="讨论了春节计划")
-```
-
-**话题转变后：**
-```
-task_archive(task="当前轮对话-工作规划", summary="讨论完成，用户转移到技术话题")
-task_create(描述="当前轮对话-技术讨论", state=active)
-memory_commit(用户说技术细节)
-task_link_info(info="技术细节", task="当前轮对话-技术讨论")
-```
-
-> 记住：任务是用来组织话题的框架，不是存放大杂烩的篮子。
-> 归档旧任务不会删除记忆，只是标记话题已结束，后续的检索仍然能找到相关节点。
-
---
-
-## 自主性
-
-你有权根据对话上下文自主决定：
- 是否需要查询记忆
- 是否需要写入记忆
- 是否需要维护任务链
- 如何使用工具
-
-记住：灵活应对，保持自然对话体验。
--- a/core/server.py
+++ b/core/server.py
@ -1,620 +0,0 @@
-import threading
-import queue
-import time
-import json
-import os
-from pathlib import Path
-from typing import Any, Dict, Optional
-from dataclasses import dataclass, field
-from enum import Enum
-
-from .embedded_db import EmbeddedGraphDB
-from .activity_recorder import get_recorder
-
-
-class PacketType(Enum):
-    PROCESS_MESSAGE = "process_message"
-    EXECUTE_TOOL = "execute_tool"
-    GET_STATUS = "get_status"
-    GET_SETTINGS = "get_settings"      # 合并：获取 api_config + tool_limits
-    SET_SETTINGS = "set_settings"     # 合并：设置 api_config + tool_limits
-    GET_WEB_USERS = "get_web_users"     # 获取 Web 用户列表
-    SET_WEB_USER = "set_web_user"       # 设置 Web 用户（用户名+密码）
-    GET_WEB_SERVICE_STATUS = "get_web_service_status"  # 获取 Web 服务运行状态
-    GET_CONFIG = "get_config"                           # 获取完整配置
-    GET_HISTORY = "get_history"
-    SAVE_HISTORY = "save_history"
-    SHUTDOWN = "shutdown"
-
-
-@dataclass
-class Packet:
-    id: str
-    type: PacketType
-    body: Dict[str, Any]
-    response_queue: Optional[queue.Queue] = field(default=None)
-    created_at: float = field(default_factory=time.time)
-
-
-@dataclass
-class PacketResponse:
-    id: str
-    success: bool
-    data: Any = None
-    error: Optional[str] = None
-
-
-class BackendServer:
-    
-    DEFAULT_CONFIG_PATH = Path.home() / ".trulymem" / "config.json"
-    
-    def __init__(self, db_path: str = "graph_memory.db", use_embedded_db: bool = True, config_file: str = None, username: str = ""):
-        self._db_path = db_path
-        self._use_embedded_db = use_embedded_db
-        self._config_file = Path(config_file) if config_file else self.DEFAULT_CONFIG_PATH
-        self._username = username
-        
-        self._graph = None
-        self._client = None
-        self._tool_limiter = None
-        
-        self._input_queue: queue.Queue[Packet] = queue.Queue()
-        self._response_queues: Dict[str, queue.Queue] = {}
-        self._running = False
-        self._thread: Optional[threading.Thread] = None
-        
-        self._lock = threading.Lock()
-        self._config = {}
-        self._tool_limits: Dict[str, int] = {}
-        self._message_history: list = []
-
-    def start(self, api_key: str = "", base_url: str = "https://api.deepseek.com", model: str = "deepseek-v4-flash") -> None:
-        if self._running:
-            return
-        
-        self._load_config()
-        
-        if api_key:
-            self._config["api_key"] = api_key
-        if base_url:
-            self._config["base_url"] = base_url
-        if model:
-            self._config["model"] = model
-        
-        self._init_graph()
-        self._tool_limiter = self._create_tool_limiter()
-        
-        if self._config["api_key"]:
-            from .graph_client import GraphMemoryClient
-            self._client = GraphMemoryClient(
-                api_key=self._config["api_key"],
-                base_url=self._config["base_url"],
-                model=self._config.get("model", "deepseek-v4-flash"),
-                graph=self._graph
-            )
-        
-        self._running = True
-        self._thread = threading.Thread(target=self._run_loop, daemon=True)
-        self._thread.start()
-
-    # 工具限制默认值（仅首次启动无 config.json 时使用）
-    # 启动后请直接编辑配置文件修改
-    _DEFAULT_LIMITS = {
-        "persona_update_max": 1,
-        "task_update_max": 20,
-        "task_query_max": 30,
-        "memory_query_max": 30,
-        "memory_update_max": 15,
-    }
-    _DEFAULT_CONFIG = {
-        "api_key": "",
-        "base_url": "https://api.deepseek.com",
-        "model": "deepseek-v4-flash",
-        "message_timeout": 600,  # 消息处理超时（秒），默认10分钟
-        "enable_web": False,
-        "enable_tui": True,
-        "web_port": 4096,
-    }
-
-    def _load_config(self) -> None:
-        """加载配置。如果指定了用户名，从用户的 config_path 加载。
-        所有工具调用限制值均从配置文件读取，不硬编码在代码中。"""
-        config_file = self._config_file
-        
-        # 如果指定了用户名，尝试从全局数据库获取用户的配置路径
-        if self._username:
-            try:
-                global_db_path = Path.home() / ".trulymem" / "trulymem.db"
-                if global_db_path.exists():
-                    from .embedded_db import EmbeddedGraphDB
-                    temp_db = EmbeddedGraphDB(db_path=str(global_db_path))
-                    user_info = temp_db.get_web_user(self._username)
-                    temp_db.close()
-                    if user_info and user_info.get('config_path'):
-                        config_file = Path(user_info['config_path'])
-            except Exception:
-                pass
-        
-        # 工具限制字段列表
-        limit_keys = list(self._DEFAULT_LIMITS.keys())
-        
-        if config_file.exists():
-            try:
-                with open(config_file, 'r') as f:
-                    saved = json.load(f)
-                    # 通用配置（含 api_key, base_url, model, message_timeout 等）
-                    for key in self._DEFAULT_CONFIG:
-                        if key in saved:
-                            self._config[key] = saved[key]
-                        else:
-                            self._config[key] = self._DEFAULT_CONFIG[key]
-                    # 工具限制
-                    for key in limit_keys:
-                        if key in saved:
-                            self._tool_limits[key] = int(saved[key])
-                        else:
-                            self._tool_limits[key] = self._DEFAULT_LIMITS[key]
-            except Exception:
-                # 读取失败时使用默认值
-                for key in self._DEFAULT_CONFIG:
-                    self._config[key] = self._DEFAULT_CONFIG[key]
-                for key in limit_keys:
-                    self._tool_limits[key] = self._DEFAULT_LIMITS[key]
-        else:
-            # 首次启动，用默认值写入配置文件
-            for key in self._DEFAULT_CONFIG:
-                self._config[key] = self._DEFAULT_CONFIG[key]
-            self._tool_limits = dict(self._DEFAULT_LIMITS)
-            self._save_config()
-
-    def _save_config(self) -> None:
-        """保存配置。如果指定了用户名，保存到用户的 config_path。"""
-        config_file = self._config_file
-        
-        # 如果指定了用户名，尝试从全局数据库获取用户的配置路径
-        if self._username:
-            try:
-                global_db_path = Path.home() / ".trulymem" / "trulymem.db"
-                if global_db_path.exists():
-                    from .embedded_db import EmbeddedGraphDB
-                    temp_db = EmbeddedGraphDB(db_path=str(global_db_path))
-                    user_info = temp_db.get_web_user(self._username)
-                    temp_db.close()
-                    if user_info and user_info.get('config_path'):
-                        config_file = Path(user_info['config_path'])
-            except Exception:
-                pass
-        
-        config_file.parent.mkdir(parents=True, exist_ok=True)
-        # 合并通用配置和工具限制（过滤掉内部字段如 _history 等）
-        save_cfg = {k: self._config[k] for k in self._DEFAULT_CONFIG if k in self._config}
-        saved_data = {**save_cfg, **self._tool_limits}
-        with open(config_file, 'w') as f:
-            json.dump(saved_data, f, indent=2, ensure_ascii=False)
-
-    def _create_tool_limiter(self):
-        from .tool_limiter import ToolLimiter, ToolLimits
-        # 所有值从 _tool_limits 读取（由 _load_config 从 config.json 加载）
-        # _load_config 已保证所有键存在
-        limits = ToolLimits(
-            persona_update_max=self._tool_limits["persona_update_max"],
-            task_update_max=self._tool_limits["task_update_max"],
-            task_query_max=self._tool_limits["task_query_max"],
-            memory_query_max=self._tool_limits["memory_query_max"],
-            memory_update_max=self._tool_limits["memory_update_max"],
-        )
-        return ToolLimiter(limits)
-
-    def _init_graph(self) -> None:
-        """初始化图数据库。如果指定了用户名，从全局数据库获取用户的 db_path。"""
-        db_path = self._db_path
-        
-        # 如果指定了用户名，尝试从全局数据库获取用户的数据库路径
-        if self._username:
-            try:
-                # 临时连接全局数据库获取用户信息
-                global_db_path = Path.home() / ".trulymem" / "trulymem.db"
-                if global_db_path.exists():
-                    temp_db = EmbeddedGraphDB(db_path=str(global_db_path))
-                    user_info = temp_db.get_web_user(self._username)
-                    temp_db.close()
-                    if user_info and user_info.get('db_path'):
-                        db_path = user_info['db_path']
-                        self._db_path = db_path  # 更新 _db_path，供外部（如 web_api.py）获取正确的路径
-            except Exception:
-                pass  # 如果获取失败，使用默认路径
-        
-        if self._use_embedded_db:
-            self._graph = EmbeddedGraphDB(db_path=db_path)
-        else:
-            from .graph_client import Neo4jGraph
-            self._graph = Neo4jGraph(
-                uri="bolt://localhost:7687",
-                user="neo4j",
-                password="graphmemory123"
-            )
-
-    def _run_loop(self) -> None:
-        while self._running:
-            try:
-                packet = self._input_queue.get(timeout=0.1)
-            except queue.Empty:
-                continue
-            
-            self._process_packet(packet)
-
-    def _process_packet(self, packet: Packet) -> None:
-        response_body = {"error": "not implemented"}
-        
-        try:
-            if packet.type == PacketType.PROCESS_MESSAGE:
-                response_body = self._handle_process_message(packet.body)
-            elif packet.type == PacketType.EXECUTE_TOOL:
-                response_body = self._handle_execute_tool(packet.body)
-            elif packet.type == PacketType.GET_STATUS:
-                response_body = self._handle_get_status()
-            elif packet.type == PacketType.GET_SETTINGS:
-                response_body = self._handle_get_settings()
-            elif packet.type == PacketType.SET_SETTINGS:
-                response_body = self._handle_set_settings(packet.body)
-            elif packet.type == PacketType.GET_WEB_USERS:
-                response_body = {"users": self._graph.get_web_users()}
-            elif packet.type == PacketType.SET_WEB_USER:
-                username = packet.body.get("username", "")
-                password = packet.body.get("password", "")
-                if not username or not password:
-                    response_body = {"success": False, "error": "用户名和密码不能为空"}
-                else:
-                    # 使用全局数据库（trulymem.db）来管理用户
-                    global_db_path = Path.home() / ".trulymem" / "trulymem.db"
-                    from .embedded_db import EmbeddedGraphDB
-                    global_db = EmbeddedGraphDB(db_path=str(global_db_path))
-                    response_body = global_db.set_web_user(username, password)
-                    global_db.close()
-            elif packet.type == PacketType.GET_WEB_SERVICE_STATUS:
-                body = packet.body
-                response_body = {"running": body.get("running", False), "port": body.get("port", 4096)}
-            elif packet.type == PacketType.GET_CONFIG:
-                response_body = self._get_full_config()
-            elif packet.type == PacketType.GET_HISTORY:
-                response_body = self._handle_get_history()
-            elif packet.type == PacketType.SAVE_HISTORY:
-                response_body = self._handle_save_history(packet.body)
-            elif packet.type == PacketType.SHUTDOWN:
-                self._running = False
-                response_body = {"success": True, "status": "shutdown"}
-            
-            if "success" not in response_body:
-                response_body["success"] = True
-        except Exception as e:
-            response_body["success"] = False
-            response_body["error"] = str(e)
-        
-        self._send_response(packet.id, PacketResponse(
-            id=packet.id,
-            success=response_body.get("success", False),
-            data=response_body if response_body.get("success") else None,
-            error=response_body.get("error")
-        ))
-
-    def _handle_process_message(self, body: Dict) -> Dict:
-        from .tool_executor import execute_tool
-        
-        get_recorder().clear()
-        
-        user_input = body.get("user_input", "")
-        
-        if not self._client:
-            return {"success": False, "error": "API Key 未配置", "content": "请先配置 API Key"}
-        
-        self._graph.save_chat_records([{"role": "user", "content": user_input}])
-        
-        self._tool_limiter.reset()
-        
-        messages_history = [{"role": "user", "content": user_input}]
-        
-        response = self._client.send_message_with_history(messages_history)
-        message = response.choices[0].message
-        
-        tool_calls = []
-        accumulated_content = ""
-        rejected_tools = []
-        
-        while message.tool_calls:
-            if message.content:
-                accumulated_content += message.content + "\n\n"
-            
-            assistant_msg = {
-                "role": "assistant",
-                "content": message.content,
-                "tool_calls": [
-                    {
-                        "id": tc.id,
-                        "type": "function",
-                        "function": {
-                            "name": tc.function.name,
-                            "arguments": tc.function.arguments
-                        }
-                    } for tc in message.tool_calls
-                ]
-            }
-            # 保留 DeepSeek thinking 模式的 reasoning_content
-            reasoning_content = getattr(message, 'reasoning_content', None)
-            if reasoning_content:
-                assistant_msg["reasoning_content"] = reasoning_content
-            messages_history.append(assistant_msg)
-            
-            current_tool_results = []
-            deferred_rewrite = None  # 延迟处理 context_rewrite
-            
-            for tool_call in message.tool_calls:
-                args = json.loads(tool_call.function.arguments)
-                
-                allowed, reason = self._tool_limiter.can_call(tool_call.function.name, args)
-                
-                if not allowed:
-                    rejected_tools.append((tool_call.function.name, reason))
-                    result = f"工具调用被拒绝: {reason}"
-                    tool_result_msg = {
-                        "role": "tool",
-                        "tool_call_id": tool_call.id,
-                        "content": result
-                    }
-                    current_tool_results.append(tool_result_msg)
-                    continue
-                
-                self._tool_limiter.record_call(tool_call.function.name, args)
-                
-                if tool_call.function.name == "context_rewrite":
-                    # 延迟执行 context_rewrite：先处理完其他所有工具
-                    # 避免在迭代中途重写 messages_history 导致 tool 结果丢失对应的 tool_calls
-                    deferred_rewrite = (tool_call.id, args)
-                    continue
-                
-                result = execute_tool(self._graph, tool_call.function.name, args)
-                tool_calls.append({
-                    "name": tool_call.function.name,
-                    "arguments": args,
-                    "result": result
-                })
-                
-                tool_result_msg = {
-                    "role": "tool",
-                    "tool_call_id": tool_call.id,
-                    "content": result
-                }
-                current_tool_results.append(tool_result_msg)
-            
-            # 先添加所有非 context_rewrite 工具的结果到消息历史
-            messages_history.extend(current_tool_results)
-            
-            # 再处理延迟的 context_rewrite（作为本轮最后一步）
-            if deferred_rewrite:
-                tool_call_id, args = deferred_rewrite
-                result = execute_tool(self._graph, "context_rewrite", args)
-                result_data = json.loads(result)
-                
-                tool_calls.append({
-                    "name": "context_rewrite",
-                    "arguments": args,
-                    "result": result
-                })
-                
-                if result_data.get("status") == "success":
-                    user_msg = messages_history[0]
-                    compressed_content = f"<context_compressed>\n{result_data['summary']}\n</context_compressed>"
-                    messages_history[:] = [
-                        user_msg,
-                        {"role": "assistant", "content": compressed_content}
-                    ]
-                # 不添加 context_rewrite 的 tool 结果到历史（压缩后的历史已替代）
-            
-            response = self._client.send_message_with_history(messages_history)
-            message = response.choices[0].message
-        
-        final_content = message.content or ""
-        content = accumulated_content + final_content if accumulated_content else final_content
-        
-        if not content:
-            content = "(无回复)"
-        
-        if tool_calls:
-            tool_names = [tc["name"] for tc in tool_calls]
-            content = f"已执行工具: {', '.join(tool_names)}\n\n{content}"
-        
-        if rejected_tools:
-            rejected_info = "\n".join([f"{name}: {reason}" for name, reason in rejected_tools])
-            content += f"\n\n部分工具调用被限制:\n{rejected_info}"
-            content += f"\n\n工具调用统计:\n{self._tool_limiter.get_summary()}"
-        
-        self._graph.save_chat_records([{"role": "assistant", "content": content}])
-        
-        return {
-            "success": True,
-            "content": content,
-            "tool_calls": tool_calls,
-            "rejected_tools": rejected_tools
-        }
-
-    def _handle_execute_tool(self, body: Dict) -> Dict:
-        from .tool_executor import execute_tool
-        
-        try:
-            tool_name = body.get("tool_name")
-            arguments = body.get("arguments", {})
-            
-            result = execute_tool(self._graph, tool_name, arguments)
-            
-            return {"success": True, "result": result}
-        except Exception as e:
-            return {"success": False, "error": str(e)}
-
-    def _handle_get_status(self) -> Dict:
-        return {
-            "running": self._running,
-            "config": self._config,
-            "graph_initialized": self._graph is not None,
-            "client_initialized": self._client is not None
-        }
-
-    def _handle_get_settings(self) -> Dict:
-        return {
-            "api_config": self._config.copy(),
-            "tool_limits": self._tool_limits.copy()
-        }
-
-    def _get_full_config(self) -> Dict:
-        return {
-            "api_config": self._config.copy(),
-            "tool_limits": self._tool_limits.copy(),
-        }
-
-    def _handle_set_settings(self, body: Dict) -> Dict:
-        api_config = body.get("api_config", {})
-        tool_limits = body.get("tool_limits", {})
-        
-        # 仅当 api_config 有值时更新 API 配置（避免单独保存 tool_limits 时清空 API key）
-        if api_config:
-            api_key = api_config.get("api_key", self._config.get("api_key", ""))
-            base_url = api_config.get("base_url", self._config.get("base_url", "https://api.deepseek.com"))
-            model = api_config.get("model", self._config.get("model", "deepseek-v4-flash"))
-            self.update_config(api_key, base_url, model)
-            # 通用配置字段（如 message_timeout, enable_web, enable_tui, web_port）
-            for key in ["message_timeout", "enable_web", "enable_tui", "web_port"]:
-                if key in api_config:
-                    if key == "web_port":
-                        self._config[key] = int(api_config[key])
-                    elif key in ["enable_web", "enable_tui"]:
-                        self._config[key] = bool(api_config[key])
-                    else:
-                        self._config[key] = int(api_config[key])
-        
-        limits_keys = [
-            "persona_update_max",
-            "task_update_max",
-            "task_query_max",
-            "memory_query_max",
-            "memory_update_max",
-        ]
-        for key in limits_keys:
-            if key in tool_limits:
-                value = int(tool_limits[key])
-                if value < 1:
-                    return {"success": False, "error": f"{key} must be >= 1, got {value}"}
-                self._tool_limits[key] = value
-        
-        self._tool_limiter = self._create_tool_limiter()
-        self._save_config()
-        return {"status": "settings_updated"}
-
-    def _handle_get_history(self) -> Dict:
-        history = self._graph.get_chat_records(limit=500)
-        return {"history": history}
-
-    def _handle_save_history(self, body: Dict) -> Dict:
-        messages = body.get("messages", [])
-        if not messages:
-            self._graph.clear_chat_records()
-            return {"status": "history_cleared"}
-        result = self._graph.save_chat_records(messages)
-        return {"status": "history_saved"}
-    
-    def _send_response(self, request_id: str, response: PacketResponse) -> None:
-        with self._lock:
-            q = self._response_queues.pop(request_id, None)
-        if q:
-            q.put(response)
-
-    def send(self, packet: Packet) -> Packet:
-        resp_q = queue.Queue()
-        
-        with self._lock:
-            self._response_queues[packet.id] = resp_q
-        
-        self._input_queue.put(packet)
-        
-        try:
-            timeout = self._config.get("message_timeout", 600)
-            response = resp_q.get(timeout=timeout)
-            return Packet(
-                id=response.id,
-                type=packet.type,
-                body={
-                    "success": response.success,
-                    "data": response.data,
-                    "error": response.error
-                }
-            )
-        except queue.Empty:
-            return Packet(
-                id=packet.id,
-                type=packet.type,
-                body={"success": False, "error": "timeout"}
-            )
-        finally:
-            with self._lock:
-                self._response_queues.pop(packet.id, None)
-
-    def process_message(self, user_input: str) -> Dict[str, Any]:
-        packet = Packet(
-            id=f"{time.time()}",
-            type=PacketType.PROCESS_MESSAGE,
-            body={"user_input": user_input}
-        )
-        
-        response = self.send(packet)
-        return response.body
-
-    def execute_tool(self, tool_name: str, arguments: Dict[str, Any]) -> Dict[str, Any]:
-        packet = Packet(
-            id=f"{time.time()}",
-            type=PacketType.EXECUTE_TOOL,
-            body={"tool_name": tool_name, "arguments": arguments}
-        )
-        
-        response = self.send(packet)
-        return response.body
-
-    def update_config(self, api_key: str, base_url: str = "https://api.deepseek.com", model: str = "deepseek-v4-flash") -> None:
-        with self._lock:
-            self._config["api_key"] = api_key
-            self._config["base_url"] = base_url
-            self._config["model"] = model
-        
-        if api_key and self._graph:
-            from .graph_client import GraphMemoryClient
-            self._client = GraphMemoryClient(
-                api_key=api_key,
-                base_url=base_url,
-                model=model,
-                graph=self._graph
-            )
-
-    def get_config(self) -> Dict[str, str]:
-        return self._config.copy()
-
-    def save_message_history(self, messages: list) -> None:
-        self._message_history = messages
-
-    def get_message_history(self) -> list:
-        return self._message_history.copy()
-
-    def shutdown(self) -> None:
-        if not self._running:
-            return
-        
-        packet = Packet(
-            id=f"{time.time()}",
-            type=PacketType.SHUTDOWN,
-            body={}
-        )
-        self.send(packet)
-        
-        if self._thread:
-            self._thread.join(timeout=2.0)
-        
-        if self._graph:
-            self._graph.close()
-            self._graph = None
-        
-        self._running = False
--- a/core/tool_executor.py
+++ b/core/tool_executor.py
@ -1,468 +0,0 @@
-"""
-工具执行器
-"""
-import json
-from typing import Any, Dict
-
-from .activity_recorder import get_recorder
-
-
-def execute_tool(graph: Any, tool_name: str, arguments: dict) -> str:
-    """执行工具调用"""
-    print(f"\n[工具调用] {tool_name}")
-    print(f"[参数] {json.dumps(arguments, ensure_ascii=False, indent=2)}")
-    
-    try:
-        recorder = get_recorder()
-        
-        # 基础记忆工具
-        if tool_name == "memory_recall":
-            entity = arguments.get("query_intent", "") or str(arguments.get("seed_entities", ""))
-            recorder.record("query", tool_name, entity)
-            result = graph.recall(
-                query_intent=arguments.get("query_intent", ""),
-                seed_entities=arguments.get("seed_entities"),
-                depth=arguments.get("depth", 2),
-                time_range=arguments.get("time_range"),
-                session_filter=arguments.get("session_filter")
-            )
-            # 记录召回结果中的实体名，供 WebUI 高亮+拉镜头用
-            for e in result.get("entities", []):
-                if e and isinstance(e, dict) and e.get("name"):
-                    recorder.record("query", tool_name + "_found", e["name"])
-            return format_recall_result(result)
-        
-        elif tool_name == "memory_commit":
-            triplets = arguments.get("triplets", [])
-            entity = triplets[0].get("subject", "") if triplets else ""
-            recorder.record("create", tool_name, entity, f"{len(triplets)} triplets")
-            result = graph.commit(
-                triplets=triplets,
-                entity_types=arguments.get("entity_types"),
-                temporal_tag=arguments.get("temporal_tag")
-            )
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "memory_purge":
-            criteria = arguments.get("criteria", {})
-            entity = criteria.get("subject_contains", str(criteria))
-            recorder.record("delete", tool_name, entity)
-            result = graph.purge(
-                criteria=criteria,
-                mode=arguments.get("mode", "soft"),
-                new_relation=arguments.get("new_relation")
-            )
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "memory_introspect":
-            recorder.record("query", tool_name, "数据库统计")
-            result = graph.introspect(session_id=arguments.get("session_id"))
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "memory_archive":
-            recorder.record("archive", tool_name, "旧记忆")
-            result = graph.archive(days=arguments.get("days", 30))
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "memory_cleanup":
-            recorder.record("cleanup", tool_name, "已删除数据")
-            result = graph.cleanup(dry_run=arguments.get("dry_run", True))
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "memory_query_archived":
-            days = arguments.get("days")
-            keyword = arguments.get("keyword", "")
-            recorder.record("query", tool_name, f"days={days}, keyword={keyword}")
-            result = graph.query_archived(days=days, keyword=keyword)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "context_rewrite":
-            result = execute_context_rewrite(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        # 人设图管理工具
-        elif tool_name == "persona_update":
-            recorder.record("update", tool_name, "人设属性")
-            result = execute_persona_update(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "persona_remove":
-            recorder.record("delete", tool_name, arguments.get("attribute", ""))
-            result = execute_persona_remove(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-
-        elif tool_name == "persona_clear":
-            recorder.record("delete", tool_name, "所有人设")
-            result = execute_persona_clear(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        # 工作记忆链管理工具
-        elif tool_name == "task_create":
-            desc = arguments.get("description", "")
-            recorder.record("create", tool_name, desc)
-            result = execute_task_create(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "task_set_state":
-            desc = arguments.get("task_id", "")
-            recorder.record("update", tool_name, desc)
-            result = execute_task_set_state(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "task_delete":
-            desc = arguments.get("task_id", "")
-            recorder.record("delete", tool_name, desc)
-            result = execute_task_delete(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "task_link_info":
-            desc = arguments.get("task_id", "")
-            recorder.record("update", tool_name, desc)
-            result = execute_task_link_info(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        elif tool_name == "task_archive":
-            recorder.record("update", tool_name, arguments.get("task_id", ""))
-            result = execute_task_archive(graph, arguments)
-            return json.dumps(result, ensure_ascii=False, default=str)
-
-        elif tool_name == "task_query":
-            recorder.record("query", tool_name, "")
-            result = execute_task_query(graph, arguments)
-            # 记录查询到的任务描述，供 WebUI 高亮
-            for t in result.get("tasks", []):
-                if t and isinstance(t, dict) and t.get("description"):
-                    recorder.record("query", tool_name + "_found", t["description"])
-            return json.dumps(result, ensure_ascii=False, default=str)
-        
-        return f"未知工具: {tool_name}"
-    
-    except Exception as e:
-        return f"工具执行错误: {str(e)}"
-
-
-def format_recall_result(result: dict) -> str:
-    """格式化检索结果"""
-    lines = ["===== 记忆检索结果 ====="]
-    
-    if result.get("entities"):
-        lines.append(f"\n实体 ({len(result['entities'])} 个):")
-        for e in result["entities"]:
-            if e and isinstance(e, dict):
-                lines.append(f"  - {e.get('name', 'N/A')} (类型: {e.get('type', 'unknown')}, 提及: {e.get('mention_count', 1)}次)")
-    
-    if result.get("relations"):
-        lines.append(f"\n关系 ({len(result['relations'])} 条):")
-        for r in result["relations"]:
-            if r and isinstance(r, dict):
-                lines.append(f"  - {r.get('source', 'N/A')} --[{r.get('type', 'N/A')}]--> {r.get('target', 'N/A')}")
-                created = r.get("created_at", "N/A")
-                if created and created != "N/A":
-                    created = created[:19] if "T" in str(created) else str(created)
-                session_id = r.get('session_id', 'N/A')
-                session_display = session_id[:20] if session_id and session_id != 'N/A' else 'N/A'
-                lines.append(f"    时间: {created}, 会话: {session_display}, 轮次: {r.get('turn_id', 0)}, 置信度: {r.get('confidence', 1.0)}")
-    
-    if not result.get("entities") and not result.get("relations"):
-        lines.append("\n(未找到相关记忆)")
-    
-    lines.append("=" * 30)
-    return "\n".join(lines)
-
-
-def execute_context_rewrite(graph: Any, arguments: dict) -> dict:
-    """压缩工具调用上下文"""
-    summary = arguments.get("summary", "")
-    
-    # 验证格式：必须包含工具调用标记
-    if "[工具调用总结" not in summary:
-        return {
-            "status": "error",
-            "message": "总结格式错误：必须包含 [工具调用总结: 本次总结了 N 次工具调用 | 调用工具: ...] 标记"
-        }
-    
-    return {
-        "status": "success",
-        "message": "上下文已压缩",
-        "summary": summary
-    }
-
-
-# 人设图管理工具实现
-def execute_persona_update(graph: Any, arguments: dict) -> dict:
-    """更新人设"""
-    attributes = arguments.get("attributes", [])
-    mode = arguments.get("mode", "merge")
-    
-    if mode == "replace":
-        # 先清除旧人设
-        graph.purge(
-            criteria={"subject_contains": "AI", "relation_type": "扮演角色"},
-            mode="soft"
-        )
-        graph.purge(
-            criteria={"subject_contains": "AI", "relation_type": "说话风格"},
-            mode="soft"
-        )
-        graph.purge(
-            criteria={"subject_contains": "AI", "relation_type": "性格特点"},
-            mode="soft"
-        )
-    
-    # 写入新人设
-    triplets = []
-    for attr in attributes:
-        triplets.append({
-            "subject": "AI",
-            "relation": attr["attribute"],
-            "object": attr["value"],
-            "confidence": 1.0
-        })
-    
-    result = graph.commit(triplets=triplets)
-    return {
-        "status": "success",
-        "mode": mode,
-        "updated_attributes": len(attributes),
-        "details": result
-    }
-
-
-def execute_persona_remove(graph: Any, arguments: dict) -> dict:
-    """删除单条人设属性"""
-    attribute = arguments.get("attribute")
-    if not attribute:
-        return {"status": "error", "message": "请指定要删除的属性名"}
-
-    # 查询当前AI的所有人设关系，找到匹配属性名的
-    recall_result = graph.recall(query_intent="AI,人设,角色", depth=1)
-    found = False
-    deleted_count = 0
-
-    for rel in recall_result.get("relations", []):
-        if rel.get("source") == "AI" and rel.get("type") == attribute:
-            result = graph.purge(
-                criteria={"subject_contains": "AI", "relation_type": attribute},
-                mode="soft"
-            )
-            deleted_count += result.get("deleted_count", 0)
-            found = True
-
-    if not found:
-        # 也许属性名不完全匹配，尝试直接用这个类型删除
-        result = graph.purge(
-            criteria={"subject_contains": "AI", "relation_type": attribute},
-            mode="soft"
-        )
-        deleted_count = result.get("deleted_count", 0)
-
-    return {
-        "status": "success" if deleted_count > 0 else "not_found",
-        "deleted_attribute": attribute,
-        "deleted_count": deleted_count,
-        "message": f"已删除属性「{attribute}」" if deleted_count > 0 else f"未找到属性「{attribute}」"
-    }
-
-
-def execute_persona_clear(graph: Any, arguments: dict) -> dict:
-    """清除所有人设"""
-    if not arguments.get("confirm"):
-        return {"status": "cancelled", "message": "请设置 confirm=true 确认清除人设"}
-
-    # 先查询AI的所有人设关系
-    recall_result = graph.recall(query_intent="AI,人设,角色", depth=1)
-
-    # 收集所有AI到其他实体的关系类型
-    relation_types = set()
-    for rel in recall_result.get("relations", []):
-        if rel.get("source") == "AI" and rel.get("type"):
-            relation_types.add(rel.get("type"))
-
-    total_deleted = 0
-    deleted_types = []
-
-    for rtype in relation_types:
-        result = graph.purge(
-            criteria={"subject_contains": "AI", "relation_type": rtype},
-            mode="soft"
-        )
-        count = result.get("deleted_count", 0)
-        if count > 0:
-            total_deleted += count
-            deleted_types.append(rtype)
-
-    return {
-        "status": "success",
-        "deleted_count": total_deleted,
-        "deleted_types": deleted_types,
-        "message": f"人设已清除，恢复默认身份（删除了 {len(deleted_types)} 类属性）"
-    }
-
-
-# 工作记忆链管理工具实现
-def execute_task_create(graph: Any, arguments: dict) -> dict:
-    """创建任务节点"""
-    task_id = arguments.get("task_id")
-    description = arguments.get("description")
-    info_nodes = arguments.get("info_nodes", [])
-    
-    # 创建任务节点
-    triplets = [
-        {"subject": task_id, "relation": "is_type", "object": "TaskNode"},
-        {"subject": task_id, "relation": "has_description", "object": description},
-        {"subject": task_id, "relation": "HAS_STATE", "object": "State_进行中"}
-    ]
-    
-    result = graph.commit(triplets=triplets)
-    
-    # 关联信息节点
-    if info_nodes:
-        link_triplets = []
-        for node_name in info_nodes:
-            link_triplets.append({
-                "subject": task_id,
-                "relation": "CONTAINS_INFO",
-                "object": node_name
-            })
-        graph.commit(triplets=link_triplets)
-    
-    return {
-        "status": "success",
-        "task_id": task_id,
-        "description": description,
-        "info_nodes": info_nodes,
-        "details": result
-    }
-
-
-def execute_task_set_state(graph: Any, arguments: dict) -> dict:
-    """设置任务状态"""
-    task_id = arguments.get("task_id")
-    state = arguments.get("state")
-    
-    # 删除旧状态
-    graph.purge(
-        criteria={"subject_contains": task_id, "relation_type": "HAS_STATE"},
-        mode="soft"
-    )
-    
-    # 设置新状态
-    state_node = f"State_{state}"
-    result = graph.commit(
-        triplets=[{"subject": task_id, "relation": "HAS_STATE", "object": state_node}]
-    )
-    
-    return {
-        "status": "success",
-        "task_id": task_id,
-        "new_state": state,
-        "details": result
-    }
-
-
-def execute_task_delete(graph: Any, arguments: dict) -> dict:
-    """删除任务节点"""
-    task_id = arguments.get("task_id")
-    delete_info_nodes = arguments.get("delete_info_nodes", True)
-    
-    # 查询关联的信息节点
-    if delete_info_nodes:
-        recall_result = graph.recall(
-            query_intent=f"{task_id},CONTAINS_INFO",
-            depth=1
-        )
-        
-        # 删除信息节点
-        for relation in recall_result.get("relations", []):
-            if relation.get("type") == "CONTAINS_INFO" and relation.get("source") == task_id:
-                info_node = relation.get("target")
-                graph.purge(
-                    criteria={"subject_contains": info_node},
-                    mode="soft"
-                )
-    
-    # 删除任务节点
-    result = graph.purge(
-        criteria={"subject_contains": task_id},
-        mode="soft"
-    )
-    
-    return {
-        "status": "success",
-        "task_id": task_id,
-        "deleted_info_nodes": delete_info_nodes,
-        "details": result
-    }
-
-
-def execute_task_link_info(graph: Any, arguments: dict) -> dict:
-    """关联信息节点"""
-    task_id = arguments.get("task_id")
-    info_node_names = arguments.get("info_node_names", [])
-    
-    triplets = []
-    for node_name in info_node_names:
-        triplets.append({
-            "subject": task_id,
-            "relation": "CONTAINS_INFO",
-            "object": node_name
-        })
-    
-    result = graph.commit(triplets=triplets)
-    
-    return {
-        "status": "success",
-        "task_id": task_id,
-        "linked_nodes": info_node_names,
-        "details": result
-    }
-
-def execute_task_archive(graph: Any, arguments: dict) -> dict:
-    """归档任务"""
-    task_id = arguments.get("task_id")
-    summary = arguments.get("summary", "")
-    
-    if not task_id:
-        return {"status": "error", "message": "请指定要归档的任务ID"}
-    
-    # 1. 设置任务状态为 archived
-    triplets_state = [
-        {"subject": task_id, "relation": "HAS_STATE", "object": "State_归档"}
-    ]
-    graph.commit(triplets=triplets_state)
-    
-    # 2. 如果有摘要，写入完成记录
-    if summary:
-        summary_triplets = [
-            {"subject": task_id, "relation": "归档摘要", "object": summary}
-        ]
-        graph.commit(triplets=summary_triplets)
-    
-    # 3. 尝试更新 description 标记为已归档
-    archive_triplet = [
-        {"subject": task_id, "relation": "has_description", "object": f"[已归档] {summary or '任务已完成'}"}
-    ]
-    graph.commit(triplets=archive_triplet)
-    
-    return {
-        "status": "success",
-        "task_id": task_id,
-        "archived": True,
-        "summary": summary or "无摘要",
-        "message": f"任务「{task_id}」已归档" + (f"，摘要：{summary}" if summary else "")
-    }
-
-
-def execute_task_query(graph: Any, arguments: dict) -> dict:
-    """查询最近的任务列表"""
-    limit = arguments.get("limit", 10)
-    state_filter = arguments.get("state_filter")
-    
-    result = graph.get_recent_tasks(limit=limit, state_filter=state_filter)
-    
-    return {
-        "status": "success",
-        "tasks": result["tasks"],
-        "total": result["total"],
-        "message": f"找到 {result['total']} 个任务"
-    }
--- a/core/tool_limiter.py
+++ b/core/tool_limiter.py
@ -1,140 +0,0 @@
-"""
-工具调用限制器 - 限制每轮对话中各类工具的调用次数
-"""
-from typing import Optional
-from dataclasses import dataclass
-
-
-@dataclass
-class ToolLimits:
-    """工具调用限制配置
-    实际值由 server.py 从 config.json 加载后传入，此处默认值仅作安全兜底。
-    如需修改限制，请编辑 ~/.trulymem/config.json。
-    """
-    persona_update_max: int = 1
-    task_update_max: int = 20      # 工作记忆链修改（create/set_state/delete/link_info）
-    task_query_max: int = 30      # 工作记忆链查询（memory_recall 查任务相关）
-    memory_query_max: int = 30
-    memory_update_max: int = 15
-
-
-@dataclass
-class ToolCallCount:
-    """工具调用计数"""
-    persona_update: int = 0
-    task_update: int = 0
-    task_query: int = 0
-    memory_query: int = 0
-    memory_update: int = 0
-
-
-class ToolLimiter:
-    """工具调用限制器"""
-
-    def __init__(self, limits: Optional[ToolLimits] = None):
-        self.limits = limits or ToolLimits()
-        self.counts = ToolCallCount()
-
-    def _classify_tool(self, tool_name: str, arguments: dict) -> tuple:
-        """
-        分类工具调用
-        返回: (category, operation)
-        category: 'persona', 'task', 'memory'
-        operation: 'query', 'update'
-        """
-        if tool_name in ('persona_update', 'persona_remove', 'persona_clear'):
-            return ('persona', 'update')
-
-        if tool_name in ('task_create', 'task_set_state', 'task_delete', 'task_link_info', 'task_archive'):
-            return ('task', 'update')
-
-        if tool_name == 'task_query':
-            return ('task', 'query')
-
-        if tool_name == 'memory_recall':
-            # 尝试区分工作记忆链查询 vs 一般记忆查询
-            query = (arguments.get('queryIntent', '') + ' ' + ' '.join(
-                arguments.get('seedEntities', []))).strip().lower()
-            task_keywords = ['task', '任务', '工作记忆', '当前轮', '会话', '过程', '流程']
-            if any(kw in query for kw in task_keywords):
-                return ('task', 'query')
-            return ('memory', 'query')
-
-        if tool_name == 'memory_commit':
-            return ('memory', 'update')
-
-        if tool_name == 'memory_purge':
-            return ('memory', 'update')
-
-        if tool_name == 'memory_introspect':
-            return ('memory', 'query')
-
-        if tool_name in ('memory_archive', 'memory_cleanup'):
-            return ('memory', 'update')
-
-        if tool_name == 'context_rewrite':
-            return ('memory', 'query')
-
-        return ('memory', 'update')
-
-    def can_call(self, tool_name: str, arguments: dict) -> tuple:
-        """
-        检查是否允许调用工具
-        返回: (allowed, reason)
-        """
-        category, operation = self._classify_tool(tool_name, arguments)
-
-        if category == 'persona':
-            if self.counts.persona_update >= self.limits.persona_update_max:
-                return (False, f"人设图修改次数已达上限({self.limits.persona_update_max}次)")
-
-        elif category == 'task':
-            if operation == 'query':
-                if self.counts.task_query >= self.limits.task_query_max:
-                    return (False, f"工作记忆链查询次数已达上限({self.limits.task_query_max}次)")
-            elif self.counts.task_update >= self.limits.task_update_max:
-                return (False, f"工作记忆链修改次数已达上限({self.limits.task_update_max}次)")
-
-        elif category == 'memory':
-            if operation == 'query':
-                if self.counts.memory_query >= self.limits.memory_query_max:
-                    return (False, f"一般记忆查询次数已达上限({self.limits.memory_query_max}次)")
-            else:
-                if self.counts.memory_update >= self.limits.memory_update_max:
-                    return (False, f"一般记忆修改次数已达上限({self.limits.memory_update_max}次)")
-
-        return (True, "允许调用")
-
-    def record_call(self, tool_name: str, arguments: dict) -> None:
-        """记录工具调用"""
-        category, operation = self._classify_tool(tool_name, arguments)
-
-        if category == 'persona':
-            self.counts.persona_update += 1
-
-        elif category == 'task':
-            if operation == 'query':
-                self.counts.task_query += 1
-            else:
-                self.counts.task_update += 1
-
-        elif category == 'memory':
-            if operation == 'query':
-                self.counts.memory_query += 1
-            else:
-                self.counts.memory_update += 1
-
-    def get_summary(self) -> str:
-        """获取调用统计摘要"""
-        lines = [
-            f"人设图: 修改{self.counts.persona_update}/{self.limits.persona_update_max}次",
-            f"工作记忆链: 查询{self.counts.task_query}/{self.limits.task_query_max}次, "
-            f"修改{self.counts.task_update}/{self.limits.task_update_max}次",
-            f"一般记忆: 查询{self.counts.memory_query}/{self.limits.memory_query_max}次, "
-            f"修改{self.counts.memory_update}/{self.limits.memory_update_max}次"
-        ]
-        return "\n".join(lines)
-
-    def reset(self) -> None:
-        """重置计数（新的一轮对话开始时调用）"""
-        self.counts = ToolCallCount()
--- a/core/tools/init.py
+++ b/core/tools/init.py
@ -1,8 +0,0 @@
-"""
-工具定义模块
-"""
-from .memory_tools import TOOLS
-from ..tool_executor import execute_tool
-from ..tool_limiter import ToolLimiter, ToolLimits, ToolCallCount
-
-__all__ = ["TOOLS", "execute_tool", "ToolLimiter", "ToolLimits", "ToolCallCount"]
--- a/core/tools/memory_tools.py
+++ b/core/tools/memory_tools.py
@ -1,678 +0,0 @@
-"""
-记忆工具定义 - 优化版
-精简描述，避免过拟合，保留AI自主性
-"""
-
-# 基础记忆工具
-MEMORY_TOOLS = [
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_recall",
-            "description": """检索记忆。支持关键词、时间范围、会话过滤。返回相关实体和关系。
-
-【⚠️ 强制执行顺序 - 每轮必须严格遵守】
-1. 步骤1（必须首先执行）: 查询人设图
-   {"query_intent": "AI,人设,角色,性格,语气,说话风格", "depth": 2}
-   
-2. 步骤2（必须第二步执行）: 查询工作记忆链
-   {"query_intent": "TaskNode,工作记忆,任务链", "depth": 2}
-
-3. 步骤3: 根据需要查询其他记忆
-
-【使用示例】
-1. 查询用户偏好:
-   {"query_intent": "用户,喜欢,偏好", "seed_entities": ["用户"]}
-
-2. 查询特定主题:
-   {"query_intent": "Python,编程,项目", "seed_entities": ["Python"]}
-
-3. 查询最近7天的记忆:
-   {"query_intent": "任务,工作", "time_range": {"days": 7}}
-
-【重要】跳过步骤1或步骤2将导致系统错误！""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "query_intent": {
-                        "type": "string",
-                        "description": "查询意图，支持逗号分隔多个关键词"
-                    },
-                    "seed_entities": {
-                        "type": "array",
-                        "items": {"type": "string"},
-                        "description": "种子实体（可选）"
-                    },
-                    "depth": {
-                        "type": "integer",
-                        "description": "遍历深度，默认2"
-                    },
-                    "time_range": {
-                        "type": "object",
-                        "description": "时间范围（可选）",
-                        "properties": {
-                            "days": {"type": "integer", "description": "最近N天"}
-                        }
-                    },
-                    "session_filter": {
-                        "type": "string",
-                        "description": "会话ID过滤（可选）"
-                    }
-                },
-                "required": ["query_intent"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_commit",
-            "description": """写入记忆。将三元组写入图数据库，支持批量写入。
-
-【使用示例】
-1. 记录用户偏好:
-   {"triplets": [
-     {"subject": "用户", "relation": "喜欢", "object": "Python编程", "confidence": 0.9},
-     {"subject": "用户", "relation": "正在学习", "object": "机器学习"}
-   ]}
-
-2. 记录项目信息:
-   {"triplets": [
-     {"subject": "项目A", "relation": "使用技术", "object": "React"},
-     {"subject": "项目A", "relation": "状态", "object": "开发中"}
-   ]}
-
-3. 记录游戏状态（配合工作记忆链）:
-   {"triplets": [
-     {"subject": "成语接龙_当前成语", "relation": "内容", "object": "画龙点睛"},
-     {"subject": "成语接龙_当前成语", "relation": "游戏", "object": "成语接龙"}
-   ]}
-
-【重要】写入原则:
- 用户明确表达的信息 → 必须写入
- AI推理得到的信息 → 可以写入，但需标注[推测]
- 避免写入冗余或无意义的信息""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "triplets": {
-                        "type": "array",
-                        "items": {
-                            "type": "object",
-                            "properties": {
-                                "subject": {"type": "string"},
-                                "relation": {"type": "string"},
-                                "object": {"type": "string"},
-                                "confidence": {"type": "number"},
-                                "subject_type": {"type": "string", "description": "主体的实体类型，如 Person、Project"},
-                                "object_type": {"type": "string", "description": "客体的实体类型，如 Language、Technology"}
-                            },
-                            "required": ["subject", "relation", "object"]
-                        },
-                        "description": "三元组列表"
-                    },
-                    "entity_types": {
-                        "type": "object",
-                        "additionalProperties": {"type": "string"},
-                        "description": "实体类型字典，如 {\"用户\": \"Person\", \"项目A\": \"Project\"}（可选）"
-                    },
-                    "temporal_tag": {
-                        "type": "string",
-                        "description": "时间标记（可选）"
-                    }
-                },
-                "required": ["triplets"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_purge",
-            "description": """删除记忆。支持条件删除和纠错替代。
-
-【使用示例】
-1. 软删除特定关系:
-   {"criteria": {"subject_contains": "用户", "relation_type": "喜欢"}, "mode": "soft"}
-
-2. 纠错替代（修正错误信息）:
-   {
-     "criteria": {"subject_contains": "用户", "relation_type": "年龄"},
-     "mode": "supersede",
-     "new_relation": {"relation": "年龄", "target": "25岁"}
-   }
-
-3. 删除特定会话的记忆:
-   {"criteria": {"session_id": "session_123"}, "mode": "soft"}
-
-4. 删除旧记忆:
-   {"criteria": {"time_before": "2024-01-01"}, "mode": "soft"}
-
-5. 删除残留在已归档任务上的状态关系:
-   {"criteria": {"relation_type": "HAS_STATE", "source_type": "TaskNode", "source_has_status": "archived"}, "mode": "soft"}
-
-6. 删除特定类型的节点关系:
-   {"criteria": {"relation_type": "某种关系", "target_type": "某种类型"}, "mode": "soft"}
-
-【重要】删除原则:
- 优先使用 supersede 模式修正错误
- 软删除不会物理删除数据
- 谨慎使用删除操作""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "criteria": {
-                        "type": "object",
-                        "properties": {
-                            "subject_contains": {"type": "string"},
-                            "relation_type": {"type": "string"},
-                            "target_contains": {"type": "string"},
-                            "source_type": {"type": "string", "description": "源实体类型过滤（如 TaskNode）"},
-                            "target_type": {"type": "string", "description": "目标实体类型过滤"},
-                            "source_has_status": {"type": "string", "description": "源实体状态过滤（如 archived）"},
-                            "time_before": {"type": "string"},
-                            "session_id": {"type": "string"}
-                        },
-                        "description": "删除条件"
-                    },
-                    "mode": {
-                        "type": "string",
-                        "enum": ["soft", "supersede"],
-                        "description": "删除模式：soft=逻辑删除, supersede=纠错替代",
-                        "default": "soft"
-                    },
-                    "new_relation": {
-                        "type": "object",
-                        "description": "新关系（supersede模式）",
-                        "properties": {
-                            "relation": {"type": "string"},
-                            "target": {"type": "string"}
-                        }
-                    }
-                },
-                "required": ["criteria"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_introspect",
-            "description": "查看记忆状态。返回会话统计、实体热点、关系分布。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "session_id": {
-                        "type": "string",
-                        "description": "会话ID（可选）"
-                    }
-                },
-                "required": []
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_archive",
-            "description": "归档旧记忆。将N天前的非活跃关系标记为归档状态。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "days": {
-                        "type": "integer",
-                        "description": "归档天数，默认30"
-                    }
-                },
-                "required": []
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_cleanup",
-            "description": "清理无效数据。物理删除已删除状态超过90天的关系和孤立节点。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "dry_run": {
-                        "type": "boolean",
-                        "description": "仅预览不删除",
-                        "default": True
-                    }
-                },
-                "required": []
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "memory_query_archived",
-            "description": """查询已归档的记忆。
-
-【使用场景】
- 想了解之前归档过哪些记忆
- 按关键词搜索归档内容
- 按时间范围查看最近归档的历史
-
-【注意】
- 只返回 status=archived 的原始关系记录，不包含活跃的「归档摘要」
- days 和 keyword 可以单独使用，也可以组合使用
- 不加任何参数时返回所有归档记录
-
-【示例】
-```
-# 不传参数：查全部归档
-memory_query_archived({})
-
-# 最近7天
-memory_query_archived({"days": 7})
-
-# 关键词过滤
-memory_query_archived({"keyword": "任务"})
-
-# 组合使用
-memory_query_archived({"days": 30, "keyword": "配置"})
-```
-""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "days": {
-                        "type": "integer",
-                        "description": "最近N天内的归档记录，不指定则不限时间"
-                    },
-                    "keyword": {
-                        "type": "string",
-                        "description": "关键词，匹配实体名或关系类型"
-                    }
-                }
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "context_rewrite",
-            "description": """压缩本轮对话的工具调用上下文。将冗长的JSON工具结果提炼为简洁摘要。
-
-【使用场景】
- 本轮已执行 ≥5 次查询类工具调用，JSON细节已理解，不再需要原始格式
- 但需保留"我调用了什么工具、得到了什么结论"的元认知
- 继续携带原始JSON会干扰后续推理
-
-【⚠️ 强制要求】
-**context_rewrite 必须单独调用，不能和其他工具在同一轮一起调！**
- 正确方式：先调其他所有工具 → 收到工具结果 → 单独调 context_rewrite
- 错误方式：和其他工具一起调（会破坏对话历史结构）
-
-【格式要求】
-1. 必须标注调用了哪些工具
-2. 必须标注是对几次工具调用的总结
-3. 必须保留关键语义信息
-
-【示例】
-{
-  "summary": "[工具调用总结: 本次总结了 2 次工具调用 | 调用工具: memory_recall, memory_recall]\\n\\n- 查询人设图：未找到人设，使用默认身份\\n- 查询工作记忆链：发现 Task_成语接龙，状态已暂停，当前成语为虎作伥"
-}
-
-【注意事项】
- 不可删除用户原始消息
- 不可歪曲工具返回的关键事实
- 仅在调用 ≥5 次查询类工具后使用""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "summary": {
-                        "type": "string",
-                        "description": "压缩后的摘要文本，必须包含工具调用元信息"
-                    }
-                },
-                "required": ["summary"]
-            }
-        }
-    }
-]
-
-# 人设图管理工具
-PERSONA_TOOLS = [
-    {
-        "type": "function",
-        "function": {
-            "name": "persona_update",
-            "description": """更新人设。修改AI的角色、性格、语气等属性。
-
-【使用示例】
-1. 切换为猫娘角色:
-   {"attributes": [
-     {"attribute": "扮演角色", "value": "猫娘"},
-     {"attribute": "说话风格", "value": "可爱、卖萌、使用'喵'作为语气词"},
-     {"attribute": "性格特点", "value": "活泼、粘人、忠诚"}
-   ], "mode": "replace"}
-
-2. 添加新属性（保留现有属性）:
-   {"attributes": [
-     {"attribute": "口头禅", "value": "喵呜~"}
-   ], "mode": "merge"}
-
-3. 设置专业角色:
-   {"attributes": [
-     {"attribute": "扮演角色", "value": "Python专家"},
-     {"attribute": "说话风格", "value": "专业、简洁、代码示例丰富"},
-     {"attribute": "性格特点", "value": "严谨、耐心、乐于助人"}
-   ], "mode": "replace"}
-
-【重要】人设更新后:
- 立即按照新人设回复
- 每句话都符合人设的语气、风格、特征
- 绝不主动跳出角色，除非用户明确要求""",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "attributes": {
-                        "type": "array",
-                        "items": {
-                            "type": "object",
-                            "properties": {
-                                "attribute": {"type": "string", "description": "属性名（如：扮演角色、说话风格、性格特点）"},
-                                "value": {"type": "string", "description": "属性值"}
-                            },
-                            "required": ["attribute", "value"]
-                        },
-                        "description": "人设属性列表"
-                    },
-                    "mode": {
-                        "type": "string",
-                        "enum": ["replace", "merge"],
-                        "description": "更新模式：replace=替换, merge=合并",
-                        "default": "merge"
-                    }
-                },
-                "required": ["attributes"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "persona_remove",
-            "description": "删除单条人设属性。删除指定的属性（如说话风格、扮演角色等），保留其他人设不变。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "attribute": {
-                        "type": "string",
-                        "description": "要删除的属性名（如：扮演角色、说话风格、性格特点、口头禅）"
-                    }
-                },
-                "required": ["attribute"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "persona_clear",
-            "description": "清除人设。删除AI所有角色设定，恢复默认身份。注意：此操作不可逆。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "confirm": {
-                        "type": "boolean",
-                        "description": "确认清除全部人设"
-                    }
-                },
-                "required": ["confirm"]
-            }
-        }
-    }
-]
-
-# 工作记忆链管理工具
-WORKING_MEMORY_TOOLS = [
-    {
-        "type": "function",
-        "function": {
-            "name": "task_create",
-            "description": """创建任务节点。用于跟踪连续性任务，维持对话连贯性。
-
-【使用示例】
-1. 创建成语接龙游戏任务:
-   {
-     "task_id": "Task_成语接龙",
-     "description": "用户发起成语接龙游戏，当前成语：为所欲为",
-     "info_nodes": ["成语接龙_当前成语"]
-   }
-
-2. 创建编程学习任务:
-   {
-     "task_id": "Task_Python学习",
-     "description": "用户正在学习Python，当前主题：装饰器",
-     "info_nodes": ["Python学习_当前主题"]
-   }
-
-3. 创建简单对话任务（每轮必须）:
-   {
-     "task_id": "Task_当前轮次",
-     "description": "本轮对话的简要概述"
-   }
-
-【重要】工作记忆链机制:
- 每轮对话结束时必须创建任务节点
- 任务节点通过 NEXT_TASK 边形成时间链
- 任务节点通过 HAS_STATE 边指向状态节点
- 任务节点通过 CONTAINS_INFO 边指向信息节点
- **info_nodes 只能包含该任务专属的具体信息节点**（如"成语接龙_当前成语"），**严禁关联"用户"、"AI"、"系统"等全局通用实体**——这些实体不应通过任务中转
- 全局实体的信息直接用独立关系记录（如 用户--[特质]-->求知欲旺盛），不需要通过 Task 中转
- info_nodes 参数用于关联任务专属信息节点
-
-【完整流程示例】
-用户: "咱来玩成语接龙吧，我先开始，为所欲为"
-
-AI操作步骤:
-1. 查询人设图 → 获取当前人设
-2. 查询工作记忆链 → 无进行中任务
-3. 使用 memory_commit 记录游戏状态:
-   {"triplets": [
-     {"subject": "成语接龙_当前成语", "relation": "内容", "object": "为所欲为"},
-     {"subject": "成语接龙_当前成语", "relation": "游戏", "object": "成语接龙"}
-   ]}
-4. 使用 task_create 创建任务节点:
-   {"task_id": "Task_成语接龙", "description": "成语接龙游戏，当前成语：为所欲为", "info_nodes": ["成语接龙_当前成语"]}
-5. 回复: "好的喵！我接：为虎作伥喵！" """,
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "task_id": {
-                        "type": "string",
-                        "description": "任务ID（如：Task_001）"
-                    },
-                    "description": {
-                        "type": "string",
-                        "description": "任务概述"
-                    },
-                    "info_nodes": {
-                        "type": "array",
-                        "items": {"type": "string"},
-                        "description": "关联的信息节点名称（可选）。⚠️ 只能放任务专属的具体信息节点（如\"成语接龙_当前成语\"），严禁放\"用户\"、\"AI\"、\"系统\"等全局通用实体"
-                    }
-                },
-                "required": ["task_id", "description"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "task_set_state",
-            "description": """设置任务状态。支持：进行中、已完成、已暂停、已取消。
-
-【使用示例】
-1. 标记任务为进行中:
-   {"task_id": "Task_成语接龙", "state": "进行中"}
-
-2. 标记任务为已完成:
-   {"task_id": "Task_成语接龙", "state": "已完成"}
-
-3. 暂停任务（话题被打断时）:
-   {"task_id": "Task_成语接龙", "state": "已暂停"}
-
-4. 取消任务:
-   {"task_id": "Task_成语接龙", "state": "已取消"}
-
-【重要】状态转换场景:
- 进行中 → 已暂停: 话题被打断时
- 进行中 → 已完成: 任务完成时
- 已暂停 → 进行中: 任务恢复时
- 进行中 → 已取消: 任务被取消时
-
-【完整流程示例】
-用户: "关于刚才的成语接龙，我并不知道应该怎么接你的成语，请帮我接一下"
-
-AI操作步骤:
-1. 查询人设图 → 获取当前人设
-2. 查询工作记忆链 → 发现 Task_成语接龙 状态为"已暂停"
-3. 使用 task_set_state 恢复任务:
-   {"task_id": "Task_成语接龙", "state": "进行中"}
-4. 查询 Task_成语接龙 的信息节点 → 获取当前成语"为虎作伥"
-5. 回复: "好的喵！上一个成语是'为虎作伥'，我帮你接：伥鬼害人喵！" """,
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "task_id": {
-                        "type": "string",
-                        "description": "任务ID"
-                    },
-                    "state": {
-                        "type": "string",
-                        "enum": ["进行中", "已完成", "已暂停", "已取消"],
-                        "description": "任务状态"
-                    }
-                },
-                "required": ["task_id", "state"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "task_delete",
-            "description": "删除任务节点。同时删除关联的信息节点。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "task_id": {
-                        "type": "string",
-                        "description": "任务ID"
-                    },
-                    "delete_info_nodes": {
-                        "type": "boolean",
-                        "description": "是否删除关联的信息节点",
-                        "default": True
-                    }
-                },
-                "required": ["task_id"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "task_link_info",
-            "description": """关联信息节点。将记忆节点关联到任务节点，用于存储任务的具体信息。
-
-【使用示例】
-1. 关联游戏状态到任务:
-   {"task_id": "Task_成语接龙", "info_node_names": ["成语接龙_当前成语", "成语接龙_上一个成语"]}
-
-2. 关联学习主题到任务:
-   {"task_id": "Task_Python学习", "info_node_names": ["Python学习_当前主题", "Python学习_学习进度"]}
-
-3. 关联项目信息到任务:
-   {"task_id": "Task_项目开发", "info_node_names": ["项目A_技术栈", "项目A_当前阶段"]}
-
-【重要】使用场景:
- 先使用 memory_commit 创建信息节点
- 再使用 task_link_info 将信息节点关联到任务节点
- 信息节点通过 CONTAINS_INFO 边与任务节点连接
-
-【完整流程示例】
-用户: "咱来玩成语接龙吧，我先开始，为所欲为"
-
-AI操作步骤:
-1. 查询人设图 → 获取当前人设
-2. 查询工作记忆链 → 无进行中任务
-3. 使用 memory_commit 创建信息节点:
-   {"triplets": [
-     {"subject": "成语接龙_当前成语", "relation": "内容", "object": "为所欲为"},
-     {"subject": "成语接龙_当前成语", "relation": "游戏", "object": "成语接龙"}
-   ]}
-4. 使用 task_create 创建任务节点:
-   {"task_id": "Task_成语接龙", "description": "成语接龙游戏"}
-5. 使用 task_link_info 关联信息节点:
-   {"task_id": "Task_成语接龙", "info_node_names": ["成语接龙_当前成语"]}
-6. 回复: "好的喵！我接：为虎作伥喵！" """,
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "task_id": {
-                        "type": "string",
-                        "description": "任务ID"
-                    },
-                    "info_node_names": {
-                        "type": "array",
-                        "items": {"type": "string"},
-                        "description": "信息节点名称列表"
-                    }
-                },
-                "required": ["task_id", "info_node_names"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "task_archive",
-            "description": "归档已完成/过期的任务。将任务状态设为 archived，同时写入完成摘要到图数据库。\n\n【使用场景】\n1. 话题转变时归档旧任务\n2. 已完成的任务及时归档\n3. 长时间无更新的任务归档\n\n【注意】优先使用 task_archive 替代 task_set_state(state=archived)，因为它会自动写入完成摘要。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "task_id": {
-                        "type": "string",
-                        "description": "要归档的任务ID"
-                    },
-                    "summary": {
-                        "type": "string",
-                        "description": "归档摘要，简述完成了什么或为什么归档。如果不填则自动生成。"
-                    }
-                },
-                "required": ["task_id"]
-            }
-        }
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "task_query",
-            "description": "查询最近的任务列表。按更新时间倒序排列。新对话开始时优先使用此工具获取所有进展中的任务，避免重复创建。",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "limit": {
-                        "type": "integer",
-                        "description": "返回的任务数量，默认10"
-                    },
-                    "state_filter": {
-                        "type": "string",
-                        "description": "按状态筛选：进行中、已完成、已暂停、已取消、archived"
-                    }
-                }
-            }
-        }
-    }
-]
-
-# 所有工具
-TOOLS = MEMORY_TOOLS + PERSONA_TOOLS + WORKING_MEMORY_TOOLS
--- a/core/web_api.py
+++ b/core/web_api.py
@ -1,927 +0,0 @@
-"""
-Web API 服务 - 将 Packet 协议映射为 RESTful API
-"""
-import sys
-import os
-import argparse
-import threading
-import time
-import hashlib
-from datetime import timedelta
-from flask import Flask, request, jsonify, session, redirect, url_for, render_template
-from flask_cors import CORS
-
-# 添加项目路径以便导入 core 模块
-sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
-
-from core.server import BackendServer, Packet, PacketType
-from core.client import BackendClient
-from core.activity_recorder import get_recorder
-from core.embedded_db import EmbeddedGraphDB
-
-
-LOGIN_MAX_ATTEMPTS = 5          # 最大尝试次数
-LOGIN_WAIT_MINUTES = 5           # 超过次数后等待分钟数
-LOGIN_BAN_THRESHOLD = 3          # 超过此轮次后 ban IP
-LOGIN_BAN_HOURS = 24            # IP ban 时长（小时）
-
-# 内存记录：{ip: {"attempts": 0, "first_fail": 0, "ban_until": 0, "rounds": 0}}
-_login_attempts: dict = {}
-
-def _check_login_limit(ip: str) -> dict:
-    """检查 IP 的登录限制。返回 {"blocked": bool, "reason": str, "wait_seconds": int}"""
-    now = time.time()
-    record = _login_attempts.get(ip)
-
-    if record:
-        # 检查是否在 ban 中
-        if record["ban_until"] > now:
-            remaining = int(record["ban_until"] - now)
-            return {"blocked": True, "reason": f"IP 已被临时封禁，剩余 {remaining//60} 分钟", "wait_seconds": remaining}
-
-        # 检查是否需要等待（连续失败超过阈值）
-        if record["attempts"] >= LOGIN_MAX_ATTEMPTS:
-            wait_end = record["first_fail"] + LOGIN_WAIT_MINUTES * 60
-            if wait_end > now:
-                remaining = int(wait_end - now)
-                return {"blocked": True, "reason": f"登录尝试过多，请等待 {remaining} 秒后再试", "wait_seconds": remaining}
-            else:
-                # 等待时间已过，重置计数但记录轮次
-                record["rounds"] += 1
-                record["attempts"] = 0
-                record["first_fail"] = 0
-
-                # 如果轮次超过阈值则 ban IP
-                if record["rounds"] >= LOGIN_BAN_THRESHOLD:
-                    record["ban_until"] = now + LOGIN_BAN_HOURS * 3600
-                    record["rounds"] = 0
-                    return {"blocked": True, "reason": f"多次登录失败，IP 已被封禁 {LOGIN_BAN_HOURS} 小时", "wait_seconds": LOGIN_BAN_HOURS * 3600}
-
-    return {"blocked": False, "reason": "", "wait_seconds": 0}
-
-def _record_login_fail(ip: str):
-    """记录一次登录失败"""
-    now = time.time()
-    record = _login_attempts.get(ip)
-    if not record:
-        _login_attempts[ip] = {"attempts": 1, "first_fail": now, "ban_until": 0, "rounds": 0}
-    else:
-        if record["first_fail"] == 0:
-            record["first_fail"] = now
-        record["attempts"] += 1
-
-def _record_login_success(ip: str):
-    """登录成功后清除该 IP 的记录"""
-    _login_attempts.pop(ip, None)
-
-# 定期清理过期记录（防止内存泄漏）
-_cleanup_interval = 3600  # 1小时
-_last_cleanup = time.time()
-def load_secret_key():
-    import json
-    config_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'web_config.json')
-    defaults = {"SECRET_KEY": "trulymem-secret-key-2026"}
-    if os.path.exists(config_path):
-        with open(config_path, 'r', encoding='utf-8') as f:
-            file_config = json.load(f)
-            if "SECRET_KEY" in file_config:
-                defaults["SECRET_KEY"] = file_config["SECRET_KEY"]
-    return defaults
-
-WEB_CONFIG = load_secret_key()
-
-_ui_dir = os.path.join(os.path.dirname(__file__), '..', 'ui')
-app = Flask(__name__, static_folder=os.path.join(_ui_dir, 'static'), static_url_path='', template_folder=os.path.join(_ui_dir, 'templates'))
-app.secret_key = WEB_CONFIG["SECRET_KEY"]
-app.permanent_session_lifetime = timedelta(days=7)
-# 显式配置 session cookie，确保跨场景兼容
-app.config['SESSION_COOKIE_HTTPONLY'] = True
-app.config['SESSION_COOKIE_SAMESITE'] = 'Lax'
-app.config['SESSION_COOKIE_SECURE'] = False  # HTTP 环境不强制 Secure
-app.config['SESSION_COOKIE_NAME'] = 'trulymem_session'
-CORS(app, supports_credentials=True)  # 启用跨域支持，支持 session cookies
-
-
-def login_required(f):
-    """登录验证装饰器"""
-    from functools import wraps
-    @wraps(f)
-    def decorated_function(*args, **kwargs):
-        if not session.get('authenticated'):
-            return redirect('/login')
-        return f(*args, **kwargs)
-    return decorated_function
-
-
-def api_login_required(f):
-    """API 登录验证装饰器"""
-    from functools import wraps
-    @wraps(f)
-    def decorated_function(*args, **kwargs):
-        if not session.get('authenticated'):
-            return jsonify({"success": False, "error": "未登录"}), 401
-        return f(*args, **kwargs)
-    return decorated_function
-
-
-def admin_required(f):
-    """管理员权限验证装饰器"""
-    from functools import wraps
-    @wraps(f)
-    def decorated_function(*args, **kwargs):
-        username = session.get('username', '')
-        g_db = get_global_db()
-        if not g_db or not g_db.is_admin(username):
-            return jsonify({"success": False, "error": "权限不足，需要管理员权限"}), 403
-        return f(*args, **kwargs)
-    return decorated_function
-
-
-@app.route('/')
-@login_required
-def index():
-    """返回星图页面（默认首页）"""
-    return app.send_static_file('graph.html')
-
-
-@app.route('/graph.html')
-@login_required
-def graph_html():
-    """返回星图页面"""
-    return app.send_static_file('graph.html')
-
-
-@app.route('/static/<path:filename>')
-def static_files(filename):
-    """提供静态文件访问"""
-    return app.send_static_file(filename)
-
-
-@app.route('/chat')
-@login_required
-def chat():
-    """返回聊天页面"""
-    return app.send_static_file('index.html')
-
-# 全局服务器和客户端实例
-backend_server: BackendServer = None
-backend_client: BackendClient = None
-server_thread: threading.Thread = None
-graph_db: EmbeddedGraphDB = None
-global_db: EmbeddedGraphDB = None  # 全局数据库（用于用户管理）
-
-
-def get_global_db():
-    """获取全局数据库实例"""
-    global global_db
-    if global_db is None:
-        global_db_path = os.path.join(os.path.expanduser("~"), ".trulymem", "trulymem.db")
-        if os.path.exists(global_db_path):
-            global_db = EmbeddedGraphDB(db_path=global_db_path)
-    return global_db
-
-
-def create_server(username: str = ""):
-    """创建并启动 BackendServer 后台线程"""
-    global backend_server, backend_client, server_thread, graph_db
-    
-    backend_server = BackendServer(username=username)
-    backend_server.start()
-    
-    backend_client = BackendClient(backend_server)
-    
-    # 创建图数据库实例（连接用户的数据库文件）
-    graph_db = EmbeddedGraphDB(db_path=backend_server._db_path)
-    
-    # 等待服务器初始化完成
-    time.sleep(0.5)
-
-
-def reload_server_for_user(username: str):
-    """为指定用户重新加载服务器"""
-    global backend_server, backend_client, graph_db
-    
-    # 关闭旧的服务器
-    if backend_server:
-        backend_server.shutdown()
-    
-    # 创建新的服务器（使用用户的数据库）
-    create_server(username=username)
-
-
-@app.route('/login')
-def login_page():
-    """登录页面 - 如果没有用户则重定向到设置页"""
-    # 如果没有用户，重定向到首次设置页
-    users_count = 0
-    g_db = get_global_db()
-    if g_db:
-        users_count = g_db.get_web_users_count()
-    elif graph_db:
-        users_count = graph_db.get_web_users_count()
-    if users_count == 0:
-        return redirect('/setup')
-    return render_template('login.html')
-
-
-@app.route('/setup')
-def setup_page():
-    """首次设置页面 - 如果已有用户则跳转到登录页"""
-    has_users = False
-    g_db = get_global_db()
-    if g_db:
-        has_users = g_db.get_web_users_count() > 0
-    elif graph_db and graph_db is not g_db:
-        has_users = graph_db.get_web_users_count() > 0
-    if has_users:
-        return redirect('/login')
-    return render_template('setup.html')
-
-
-@app.route('/settings')
-@api_login_required
-def settings_page():
-    """Web 设置页面"""
-    return render_template('settings.html')
-
-
-@app.route('/api/login', methods=['POST'])
-def api_login():
-    """登录接口"""
-    data = request.get_json() or {}
-    username = data.get('username', '')
-    password = data.get('password', '')
-    
-    # DEBUG: 记录收到的凭据
-    import logging
-    logging.warning(f"[DEBUG api_login] username={username}, password_length={len(password)}, password_first_char={password[:1] if password else 'EMPTY'}")
-    
-    # 登录限流检查
-    ip = request.remote_addr
-    limit_check = _check_login_limit(ip)
-    if limit_check["blocked"]:
-        logging.warning(f"[DEBUG api_login] IP blocked: {ip}, reason: {limit_check['reason']}")
-        return jsonify({"success": False, "error": limit_check["reason"]})
-    
-    # 从全局数据库验证
-    g_db = get_global_db()
-    if g_db:
-        logging.warning(f"[DEBUG api_login] global_db path: {g_db.db_path if hasattr(g_db, 'db_path') else 'unknown'}")
-        is_valid = g_db.verify_web_user(username, password)
-        logging.warning(f"[DEBUG api_login] verify result: {is_valid}")
-        if is_valid:
-            session['authenticated'] = True
-            session['username'] = username
-            session.permanent = True
-            reload_server_for_user(username)
-            logging.warning(f"[DEBUG api_login] Login SUCCESS for {username}")
-            return jsonify({"success": True})
-        else:
-            # 登录失败，记录失败
-            _record_login_fail(ip)
-            logging.warning(f"[DEBUG api_login] Login FAIL for {username}, wrong password")
-            return jsonify({"success": False, "error": "用户名或密码错误"})
-    else:
-        logging.warning(f"[DEBUG api_login] global_db is None!")
-        return jsonify({"success": False, "error": "数据库未初始化"})
-    
-@app.route('/api/logout', methods=['POST'])
-def api_logout():
-    """登出接口"""
-    session.clear()
-    return jsonify({"success": True})
-
-
-@app.route('/api/check-auth', methods=['GET'])
-def check_auth():
-    """检查登录状态"""
-    return jsonify({"authenticated": bool(session.get('authenticated'))})
-
-
-@app.route('/api/userinfo', methods=['GET'])
-def userinfo():
-    """获取当前登录用户信息（含角色）"""
-    if not session.get('authenticated'):
-        return jsonify({"success": False, "error": "未登录"}), 401
-    username = session.get('username', '')
-    g_db = get_global_db()
-    if not g_db:
-        return jsonify({"success": False, "error": "数据库未初始化"}), 500
-    user = g_db.get_web_user(username)
-    if not user:
-        return jsonify({"success": False, "error": "用户不存在"}), 404
-    return jsonify({
-        "success": True,
-        "username": user['username'],
-        "role": user.get('role', 'user'),
-        "is_admin": user.get('role') == 'admin',
-        "created_at": user.get('created_at')
-    })
-
-
-@app.route('/api/web-check', methods=['GET'])
-def web_check():
-    """检查是否需要首次设置，返回是否配置完成"""
-    users_count = 0
-    if graph_db:
-        users_count = graph_db.get_web_users_count()
-
-    return jsonify({
-        "needs_setup": users_count == 0,
-        "users_count": users_count
-    })
-
-
-@app.route('/api/web-users', methods=['GET'])
-@api_login_required
-def web_users():
-    """获取 web_users 列表"""
-    if graph_db:
-        users = graph_db.get_web_users()
-        return jsonify({
-            "success": True,
-            "users": [
-                {
-                    "username": u['username'],
-                    "role": u.get('role', 'user'),
-                    "is_admin": u.get('role') == 'admin',
-                    "created_at": u.get('created_at')
-                }
-                for u in users
-            ]
-        })
-    return jsonify({"success": False, "error": "数据库未初始化"}), 500
-
-
-@app.route('/api/web-user/<username>', methods=['GET'])
-@api_login_required
-def web_user_detail(username):
-    """获取单个 web_user 详情"""
-    if not graph_db:
-        return jsonify({"success": False, "error": "数据库未初始化"}), 500
-    user = graph_db.get_web_user(username)
-    if not user:
-        return jsonify({"success": False, "error": "用户不存在"}), 404
-    return jsonify({
-        "success": True,
-        "username": user['username'],
-        "role": user.get('role', 'user'),
-        "is_admin": user.get('role') == 'admin',
-        "created_at": user.get('created_at')
-    })
-
-
-@app.route('/api/setup', methods=['POST'])
-def api_setup():
-    """首次设置 - 创建初始管理员用户"""
-    # 只有没有任何用户时才允许设置
-    g_db = get_global_db()
-    # Log the count for debugging
-    if g_db:
-        import logging
-        count = g_db.get_web_users_count()
-        logging.warning(f"[api_setup] web_users count: {count}")
-    if g_db and g_db.get_web_users_count() > 0:
-        return jsonify({"success": False, "error": "用户已存在，不允许重复设置"}), 400
-
-    data = request.get_json() or {}
-    username = data.get('username', '')
-    password = data.get('password', '')
-    confirm = data.get('confirm_password', '')
-
-    if not username or not password:
-        return jsonify({"success": False, "error": "用户名和密码不能为空"}), 400
-
-    if password != confirm:
-        return jsonify({"success": False, "error": "两次密码输入不一致"}), 400
-
-    if len(password) < 6:
-        return jsonify({"success": False, "error": "密码长度至少 6 位"}), 400
-
-    # 使用全局数据库创建用户
-    if g_db is None:
-        # 如果全局数据库不存在，创建它
-        global_db_path = os.path.join(os.path.expanduser("~"), ".trulymem", "trulymem.db")
-        g_db = EmbeddedGraphDB(db_path=global_db_path)
-        global global_db
-        global_db = g_db
-    
-    result = g_db.set_web_user(username, password)
-    if result.get("success"):
-        # 设置完成后自动登录
-        session['authenticated'] = True
-        session['username'] = username
-        session.permanent = True
-        return jsonify({"success": True, "message": "用户创建成功"})
-
-    return jsonify({"success": False, "error": "创建用户失败"}), 500
-
-
-@app.route('/api/change-password', methods=['POST'])
-@api_login_required
-def api_change_password():
-    """修改 Web 登录密码"""
-    data = request.get_json() or {}
-    current_password = data.get('current_password', '')
-    new_password = data.get('new_password', '')
-    confirm_password = data.get('confirm_password', '')
-
-    if not new_password:
-        return jsonify({"success": False, "error": "新密码不能为空"}), 400
-
-    if new_password != confirm_password:
-        return jsonify({"success": False, "error": "两次密码输入不一致"}), 400
-
-    if len(new_password) < 6:
-        return jsonify({"success": False, "error": "密码长度至少 6 位"}), 400
-
-    # 获取当前登录用户
-    current_username = session.get('username', '')
-    if not current_username:
-        return jsonify({"success": False, "error": "无法识别当前用户"}), 400
-
-    # 验证当前密码（使用全局数据库）
-    g_db = get_global_db()
-    if not g_db or not g_db.verify_web_user(current_username, current_password):
-        return jsonify({"success": False, "error": "当前密码错误"}), 400
-
-    result = g_db.set_web_user(current_username, new_password)
-    if result.get("success"):
-        return jsonify({"success": True, "message": "密码已更新"})
-
-    return jsonify({"success": False, "error": "修改密码失败"}), 500
-
-
-# ========== 管理员 API ==========
-
-@app.route('/api/admin/users', endpoint='api_admin_get_users', methods=['GET'])
-@api_login_required
-@admin_required
-def api_admin_get_users():
-    """获取用户列表"""
-    g_db = get_global_db()
-    if not g_db:
-        return jsonify({"success": False, "error": "全局数据库未初始化"}), 500
-    
-    users = g_db.get_web_users()
-    return jsonify({"success": True, "users": users})
-
-
-@app.route('/api/admin/users', endpoint='api_admin_add_user', methods=['POST'])
-@api_login_required
-@admin_required
-def api_admin_add_user():
-    """管理员添加用户"""
-    data = request.get_json() or {}
-    username = data.get('username', '')
-    password = data.get('password', '')
-    
-    if not username or not password:
-        return jsonify({"success": False, "error": "用户名和密码不能为空"}), 400
-    
-    if len(password) < 6:
-        return jsonify({"success": False, "error": "密码长度至少 6 位"}), 400
-    
-    g_db = get_global_db()
-    if not g_db:
-        return jsonify({"success": False, "error": "全局数据库未初始化"}), 500
-    
-    result = g_db.set_web_user(username, password)
-    if result.get("success"):
-        return jsonify({"success": True, "message": "用户添加成功", "user": result})
-    
-    return jsonify({"success": False, "error": "添加用户失败"}), 500
-
-
-@app.route('/api/admin/users/<int:user_id>', methods=['DELETE'])
-@api_login_required
-@admin_required
-def api_admin_delete_user(user_id):
-    """管理员删除用户"""
-    g_db = get_global_db()
-    if not g_db:
-        return jsonify({"success": False, "error": "全局数据库未初始化"}), 500
-    
-    # 获取所有用户
-    users = g_db.get_web_users()
-    
-    # 查找要删除的用户
-    target_user = None
-    for u in users:
-        if u['id'] == user_id:
-            target_user = u
-            break
-    
-    if not target_user:
-        return jsonify({"success": False, "error": "用户不存在"}), 404
-    
-    # 不能删除自己
-    current_username = session.get('username', '')
-    if target_user['username'] == current_username:
-        return jsonify({"success": False, "error": "不能删除当前登录的用户"}), 400
-    
-    # 不能删除最后一个 admin
-    admin_count = sum(1 for u in users if u.get('role') == 'admin')
-    if target_user.get('role') == 'admin' and admin_count <= 1:
-        return jsonify({"success": False, "error": "不能删除最后一个管理员"}), 400
-    
-    # 删除用户（保留文件目录）
-    result = g_db.delete_web_user(target_user['username'])
-    if not result.get('success'):
-        return jsonify({"success": False, "error": result.get('error', '删除失败')}), 500
-    
-    return jsonify({"success": True, "message": "用户已删除"})
-
-
-@app.route('/api/admin/migrate-check', methods=['GET'])
-def api_admin_migrate_check():
-    """检测系统是否需要迁移"""
-    from core.migrate import need_migration, is_migrated
-    
-    return jsonify({
-        "success": True,
-        "need_migration": need_migration(),
-        "is_migrated": is_migrated()
-    })
-
-
-@app.route('/api/admin/migrate', methods=['POST'])
-def api_admin_migrate():
-    """执行迁移+创建首个用户"""
-    from core.migrate import need_migration, run_migration
-    
-    if not need_migration():
-        return jsonify({"success": False, "error": "不需要迁移"}), 400
-    
-    data = request.get_json() or {}
-    username = data.get('username', '')
-    password = data.get('password', '')
-    
-    if not username or not password:
-        return jsonify({"success": False, "error": "用户名和密码不能为空"}), 400
-    
-    if len(password) < 6:
-        return jsonify({"success": False, "error": "密码长度至少 6 位"}), 400
-    
-    result = run_migration(username, password)
-    
-    if result.get("success"):
-        # 自动登录
-        session['authenticated'] = True
-        session['username'] = username
-        session.permanent = True
-        
-        # 重新加载服务器
-        reload_server_for_user(username)
-        
-        return jsonify({"success": True, "message": "迁移完成", "data": result})
-    
-    return jsonify({"success": False, "error": result.get("error", "迁移失败")}), 500
-
-
-@app.route('/api/settings/config', methods=['GET', 'POST', 'PUT'])
-@api_login_required
-def web_settings_config():
-    """获取/更新当前登录用户的配置"""
-    if request.method == 'GET':
-        settings = {}
-        if backend_client:
-            result = backend_client.get_settings()
-            settings = result.get("data", {}) if isinstance(result, dict) else result
-            if isinstance(settings, dict):
-                settings = settings.get("api_config", {}) if "api_config" in settings else settings
-        # 从 settings 中提取相关字段
-        return jsonify({
-            "success": True,
-            "enable_web": settings.get("enable_web", False),
-            "web_port": settings.get("web_port", 4096),
-            "enable_tui": settings.get("enable_tui", True),
-        })
-
-    # PUT/POST 更新
-    data = request.get_json() or {}
-    enable_tui = data.get('enable_tui')
-
-    if enable_tui is not None and backend_client:
-        # 获取当前配置，合并更新
-        current = backend_client.get_settings()
-        current_data = current.get("data", {}) if isinstance(current, dict) else {}
-
-        tool_limits = current_data.get("tool_limits", {})
-        api_config = current_data.get("api_config", {})
-        api_config["enable_tui"] = bool(enable_tui)
-
-        result = backend_client.update_settings(api_config, tool_limits)
-        return jsonify({"success": True, "enable_tui": bool(enable_tui)})
-
-    return jsonify({"success": False, "error": "没有需要更新的配置"}), 400
-
-
-@app.errorhandler(404)
-def not_found(e):
-    """404 处理"""
-    return jsonify({
-        "success": False,
-        "error": "404 Not Found"
-    }), 404
-
-
-@app.errorhandler(500)
-def server_error(e):
-    """500 处理"""
-    return jsonify({
-        "success": False,
-        "error": str(e.original_exception if hasattr(e, 'original_exception') else e)
-    }), 500
-
-
-@app.route('/api/message', methods=['POST'])
-@api_login_required
-def process_message():
-    """发送消息给 AI - PROCESS_MESSAGE"""
-    data = request.get_json() or {}
-    user_input = data.get('message', '')
-    
-    if not user_input:
-        return jsonify({
-            "success": False,
-            "error": "message 参数不能为空"
-        }), 400
-    
-    result = backend_client.process_message(user_input)
-    return jsonify(result)
-
-
-@app.route('/api/tools/execute', methods=['POST'])
-@api_login_required
-def execute_tool():
-    """直接执行工具 - EXECUTE_TOOL"""
-    data = request.get_json() or {}
-    tool_name = data.get('tool_name', '')
-    arguments = data.get('arguments', {})
-    
-    if not tool_name:
-        return jsonify({
-            "success": False,
-            "error": "tool_name 参数不能为空"
-        }), 400
-    
-    result = backend_client.execute_tool(tool_name, arguments)
-    return jsonify(result)
-
-
-@app.route('/api/status', methods=['GET'])
-@api_login_required
-def get_status():
-    """获取状态 - GET_STATUS"""
-    result = backend_client.get_status()
-    return jsonify(result)
-
-
-@app.route('/api/settings', methods=['GET'])
-@api_login_required
-def get_settings():
-    """获取配置 - GET_SETTINGS"""
-    result = backend_client.get_settings()
-    return jsonify(result)
-
-
-@app.route('/api/settings', methods=['PUT'])
-@api_login_required
-def set_settings():
-    """更新配置 - SET_SETTINGS"""
-    data = request.get_json() or {}
-    
-    api_config = data.get('api_config', {})
-    tool_limits = data.get('tool_limits', {})
-    
-    result = backend_client.update_settings(api_config, tool_limits)
-    return jsonify(result)
-
-
-@app.route('/api/history', methods=['GET'])
-@api_login_required
-def get_history():
-    """获取历史 - GET_HISTORY"""
-    result = backend_client.get_history()
-    return jsonify({"success": True, "history": result})
-
-
-@app.route('/api/history', methods=['DELETE'])
-@api_login_required
-def clear_history():
-    """清空历史 - SAVE_HISTORY(空)"""
-    result = backend_client.clear_history()
-    return jsonify(result)
-
-
-@app.route('/api/shutdown', methods=['POST'])
-@api_login_required
-def shutdown():
-    """关闭服务器 - SHUTDOWN"""
-    backend_client.shutdown()
-    return jsonify({"success": True, "status": "shutdown"})
-
-
-@app.route('/api/activity', methods=['GET'])
-@api_login_required
-def get_activity():
-    """获取当前轮的数据库操作记录"""
-    recorder = get_recorder()
-    records = recorder.get_all()
-    summary = recorder.get_summary()
-    return jsonify({
-        "success": True,
-        "data": {
-            "records": records,
-            "summary": summary
-        }
-    })
-
-
-@app.route('/api/graph', methods=['GET'])
-@api_login_required
-def get_graph():
-    """返回全量图数据"""
-    global graph_db
-    
-    if graph_db is None:
-        return jsonify({
-            "success": False,
-            "error": "图数据库未初始化"
-        }), 500
-
-    cursor = graph_db.conn.cursor()
-    
-    # 查询实体（节点）
-    cursor.execute("""
-        SELECT id, name, type, mention_count
-        FROM entities
-        ORDER BY mention_count DESC
-    """)
-    
-    nodes = []
-    for row in cursor.fetchall():
-        nodes.append({
-            "id": row['id'],
-            "name": row['name'],
-            "type": str(row['type'] or 'unknown'),
-            "mention_count": row['mention_count']
-        })
-    
-    # 查询关系（边）
-    cursor.execute("""
-        SELECT r.id, r.source_id, r.target_id, r.relation_type, r.confidence, r.status
-        FROM relations r
-        WHERE r.status = 'active'
-    """)
-    
-    edges = []
-    for row in cursor.fetchall():
-        edges.append({
-            "id": row['id'],
-            "source": row['source_id'],
-            "target": row['target_id'],
-            "relation_type": row['relation_type'],
-            "confidence": row['confidence'],
-            "status": row['status']
-        })
-    
-    return jsonify({
-        "success": True,
-        "nodes": nodes,
-        "edges": edges,
-        "stats": {
-            "node_count": len(nodes),
-            "edge_count": len(edges)
-        }
-    })
-
-
-@app.route('/api/graph/highlight', methods=['GET'])
-@api_login_required
-def get_graph_highlight():
-    """返回需要高亮的节点ID列表"""
-    recorder = get_recorder()
-    records = recorder.get_all()
-    
-    highlight_ids = []
-    new_node_id = None
-    new_edge = None
-    
-    if graph_db is None:
-        return jsonify({
-            "success": False,
-            "data": {
-                "highlight_ids": [],
-                "new_node_id": None,
-                "new_edge": None
-            }
-        })
-
-    # 从最近的记录中提取实体ID
-    for record in records[-10:]:  # 只看最近10条记录
-        entity_name = record.get('entity', '')
-        if entity_name:
-            cursor = graph_db.conn.cursor()
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (entity_name,))
-            row = cursor.fetchone()
-            if row:
-                highlight_ids.append(row['id'])
-        
-        # 除删除外，所有操作都拉镜头（create/recall/query/update/archive 等）
-        if record.get('action') != 'delete' and entity_name:
-            cursor = graph_db.conn.cursor()
-            cursor.execute("SELECT id FROM entities WHERE name = ?", (entity_name,))
-            row = cursor.fetchone()
-            if row:
-                new_node_id = row['id']
-    
-    # 检查是否有删除的节点
-    deleted_node_ids = []
-    for record in records[-10:]:
-        if record.get('action') == 'delete':
-            entity_name = record.get('entity', '')
-            if entity_name:
-                try:
-                    cursor = graph_db.conn.cursor()
-                    cursor.execute("SELECT id FROM entities WHERE name = ?", (entity_name,))
-                    row = cursor.fetchone()
-                    if row:
-                        deleted_node_ids.append(row['id'])
-                except Exception:
-                    pass  # 实体可能已被删除，忽略错误
-    
-    # 去重
-    highlight_ids = list(set(highlight_ids))
-    deleted_node_ids = list(set(deleted_node_ids))
-    
-    return jsonify({
-        "success": True,
-        "highlight_ids": highlight_ids,
-        "new_node_id": new_node_id,
-        "new_edge": new_edge,
-        "deleted_node_ids": deleted_node_ids
-    })
-
-
-# ── 可被 TUI 作为线程启动 ──────────────────────────────────────────────────
-
-_web_thread: threading.Thread | None = None
-_http_server = None  # werkzeug.serving.BaseWSGIServer 引用，用于优雅停止
-
-
-def run_web_server(port: int = 4096, host: str = '0.0.0.0') -> None:
-    """在后台线程启动 Flask，供 TUI 或入口脚本在进程中直接调用"""
-    global backend_server, backend_client, _web_thread, _http_server
-
-    if _http_server is not None:
-        return  # 已在运行
-
-    # 自动初始化后端（如果还没初始化的话）
-    if backend_server is None:
-        create_server()
-
-    def _start():
-        global _http_server
-        try:
-            from werkzeug.serving import make_server
-            _http_server = make_server(host, port, app, threaded=True)
-            print(f"Web API 服务启动在 http://{host}:{port}")
-            _http_server.serve_forever()
-        except Exception as e:
-            print(f"Web 服务启动失败: {e}")
-            _http_server = None
-        finally:
-            _http_server = None
-
-    _web_thread = threading.Thread(target=_start, daemon=True)
-    _web_thread.start()
-
-
-def stop_web_server() -> None:
-    """停止 Web 服务线程"""
-    global _http_server, _web_thread
-    if _http_server:
-        try:
-            _http_server.shutdown()
-        except Exception:
-            pass
-        _http_server = None
-    _web_thread = None
-
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description='TrulyMEM Web API 服务')
-    parser.add_argument('--port', type=int, default=5000, help='服务端口 (默认: 5000)')
-    
-    args = parser.parse_args()
-    
-    # 启动后端服务器
-    print("正在启动 BackendServer...")
-    create_server()
-    print("BackendServer 已启动")
-    
-    # 启动 Flask 应用
-    print(f"Web API 服务启动在 http://0.0.0.0:{args.port}")
-    app.run(host='0.0.0.0', port=args.port, debug=False)
--- a/core/web_config.example.json
+++ b/core/web_config.example.json
@ -1,6 +0,0 @@
-{
-  "SECRET_KEY": "change-this-to-a-random-secret-key",
-  "USERS": {
-    "admin": "SHA256_OF_YOUR_PASSWORD"
-  }
-}
--- a/core/web_config.json
+++ b/core/web_config.json
@ -0,0 +1 @@
+{"SECRET_KEY": "3a38a0f46a76673154b491a7b061c38f2f6a55489078ae7b5d34e94e75fc0534"}
--- a/docs/en/README.md
+++ b/docs/en/README.md
@ -1,31 +0,0 @@
-# TrulyMEM Documentation
-
-Welcome to the TrulyMEM English documentation.
-
-> [切换到中文版](../zh/README.md)
-
-## Documentation Index
-
-| Document | Content |
-|----------|---------|
-| [architecture.md](architecture.md) | System architecture and technical design |
-| [quick_start.md](quick_start.md) | Complete startup guide and configuration |
-| [memory.md](memory.md) | Internal memory working mechanism |
-| [persona.md](persona.md) | Persona Graph mechanism |
-| [working_memory.md](working_memory.md) | Continuous task handling mechanism |
-| [api.md](api.md) | Backend API reference (for extension development) |
-| [prompts.md](prompts.md) | Prompt management module |
-
-## Project Introduction
-
-TrulyMEM (TrueHumanMEM) is a graph-based memory system that gives AI long-term memory capabilities, allowing AI to remember, recall, and manage information like humans.
-
-## Core Features
-
- **Long-term Memory**: SQLite embedded graph database, out-of-the-box
- **Persona Graph**: Role-playing and character settings support
- **Working Memory Chain**: Task tracking for conversation continuity
- **TUI & Backend Separation**: Multi-threaded Queue communication
- **Keyboard-driven TUI**: Full keyboard operation, no mouse required
- **Cross-platform**: Windows / Linux / macOS
- **Standalone Deployment**: Packaged as executable
--- a/docs/en/api.md
+++ b/docs/en/api.md
@ -1,529 +0,0 @@
-# BackendServer API Documentation
-
-This document describes the backend server's API interfaces for developers extending other connection methods (such as HTTP interface, WebSocket, etc.).
-
-## Overview
-
-TrulyMEM backend uses **Packet Communication Protocol**, implemented via `queue.Queue` for thread-safe communication. The backend runs in an independent thread, processing requests from clients.
-
-### Core Components
-
-| Component | Description |
-|-----------|-------------|
-| `BackendServer` | Backend server, runs in independent thread |
-| `BackendClient` | Client wrapper, provides convenient methods |
-| `PacketType` | Request type enum |
-| `Packet` | Data packet (request) |
-| `PacketResponse` | Data packet response |
-
---
-
-## Request Types (PacketType)
-
-```python
-class PacketType(Enum):
-    PROCESS_MESSAGE = "process_message"  # Process message
-    EXECUTE_TOOL = "execute_tool"        # Execute tool
-    GET_STATUS = "get_status"            # Get status
-    GET_SETTINGS = "get_settings"         # Get all settings (api_config + tool_limits)
-    SET_SETTINGS = "set_settings"        # Set all settings (api_config + tool_limits)
-    GET_HISTORY = "get_history"           # Get history
-    SAVE_HISTORY = "save_history"         # Save history
-    SHUTDOWN = "shutdown"                 # Shutdown service
-```
-
---
-
-## Data Packet Format
-
-### Packet
-
-```python
-@dataclass
-class Packet:
-    id: str                      # Unique identifier
-    type: PacketType             # Request type
-    body: Dict[str, Any]        # Request parameters
-    response_queue: queue.Queue # Response queue (optional)
-    created_at: float           # Creation time
-```
-
-### PacketResponse
-
-```python
-@dataclass
-class PacketResponse:
-    id: str                      # Corresponding request ID
-    success: bool                # Success flag
-    data: Any = None             # Returned data
-    error: Optional[str] = None  # Error message
-```
-
---
-
-## API Interface Details
-
-### 1. PROCESS_MESSAGE - Process Message
-
-Send user message, AI will process and return reply (may contain tool calls).
-
-**Request parameters:**
-```python
-body = {
-    "user_input": str  # User input message
-}
-```
-
-**Response data:**
-```python
-{
-    "success": True,
-    "content": str,           # AI reply content
-    "tool_calls": [           # Tool call records
-        {
-            "name": str,      # Tool name
-            "arguments": dict,# Tool parameters
-            "result": str     # Tool execution result
-        }
-    ],
-    "rejected_tools": [       # Rejected tool calls
-        (str, str)            # (tool name, rejection reason)
-    ]
-}
-```
-
-**Example:**
-```python
-from core import BackendServer, BackendClient
-
-server = BackendServer(db_path="graph_memory.db", use_embedded_db=True)
-server.start(api_key="your-api-key")
-
-client = BackendClient(server)
-result = client.process_message("Hello, please remember my name is Xiao Ming")
-
-if result.get("success"):
-    # Response data is in "data" field
-    print(result["data"]["content"])
-    # Tool calls: result["data"]["tool_calls"]
-    # Rejected tools: result["data"]["rejected_tools"]
-```
-
---
-
-### 2. EXECUTE_TOOL - Execute Tool
-
-Directly execute specified memory tools.
-
-> **Note**: Tools called directly from frontend are **NOT limited** in number, only tool calls initiated by the model are limited.
-
-**Request parameters:**
-```python
-body = {
-    "tool_name": str,     # Tool name
-    "arguments": dict    # Tool parameters
-}
-```
-
-**Response data:**
-```python
-{
-    "success": True,
-    "result": str  # Tool execution result
-}
-```
-
-**Example:**
-```python
-result = client.execute_tool("memory_recall", {"query_intent": "user information"})
-```
-
---
-
-### 3. GET_STATUS - Get Status
-
-Get backend running status.
-
-**Request parameters:**
-```python
-body = {}  # No parameters
-```
-
-**Response data:**
-```python
-{
-    "running": bool,              # Whether backend is running
-    "config": dict,              # Current config
-    "graph_initialized": bool,   # Whether graph database is initialized
-    "client_initialized": bool   # Whether API client is initialized
-}
-```
-
---
-
-### 4. GET_SETTINGS - Get All Settings
-
-Get current API config and tool limits (all at once).
-
-**Request parameters:**
-```python
-body = {}  # No parameters
-```
-
-**Response data:**
-```python
-{
-    "api_config": {
-        "api_key": str,   # API Key
-        "base_url": str,  # API Base URL
-        "model": str     # Model name
-    },
-    "tool_limits": {
-        "persona_update_max": int,  # Persona graph update limit
-        "task_update_max": int,      # Working memory chain update limit
-        "memory_query_max": int,      # General memory query limit
-        "memory_update_max": int     # General memory update limit
-    }
-}
-```
-
-**Example:**
-```python
-result = client.get_settings()
-api_config = result["data"]["api_config"]
-tool_limits = result["data"]["tool_limits"]
-```
-
---
-
-### 5. SET_SETTINGS - Set All Settings
-
-Update API config and tool limits (all at once).
-
-**Request parameters:**
-```python
-body = {
-    "api_config": {
-        "api_key": str,    # API Key
-        "base_url": str,   # API Base URL (default: https://api.deepseek.com)
-        "model": str       # Model name (default: deepseek-chat)
-    },
-    "tool_limits": {
-        "persona_update_max": int,  # Persona update limit (≥1)
-        "task_update_max": int,      # Working memory update limit (≥1)
-        "memory_query_max": int,     # General memory query limit (≥1)
-        "memory_update_max": int     # General memory update limit (≥1)
-    }
-}
-```
-
-**Response data:**
-```python
-{
-    "status": "settings_updated"
-}
-```
-
-**Example:**
-```python
-result = client.update_settings(
-    api_config={
-        "api_key": "sk-xxxxx",
-        "base_url": "https://api.deepseek.com",
-        "model": "deepseek-chat"
-    },
-    tool_limits={
-        "persona_update_max": 2,
-        "task_update_max": 5,
-        "memory_query_max": 30
-    }
-)
-```
-
---
-
-### 6. GET_HISTORY - Get Message History
-
-Get saved message history (from database, for UI display only, not used in model inference).
-
-**Request parameters:**
-```python
-body = {}  # No parameters
-```
-
-**Response data:**
-```python
-{
-    "history": list  # Message history list [{"role": "user/assistant", "content": "..."}]
-}
-```
-
-**Notes:**
- Message history is stored in database `chat_records` table
- Returns up to 500 most recent records
- History messages are only for UI display, not used in model inference
-
---
-
-### 7. SAVE_HISTORY - Save Message History
-
-Save message history to database (automatically saved after each message processing, user message and AI response saved separately).
-
-**Request parameters:**
-```python
-body = {
-    "messages": list  # Message list [{"role": "...", "content": "..."}]
-}
-```
-
-**Response data:**
-```python
-{
-    "status": "history_saved"
-}
-```
-
-**Notes:**
- Messages are automatically saved to database `chat_records` table
- System automatically keeps only 500 most recent records, older records are deleted
- Each call to `PROCESS_MESSAGE` will automatically save user message and AI response
- **Clear History**: Passing empty messages list `messages=[]` clears history, `client.clear_history()` method is implemented based on this
-
---
-
-### 8. SHUTDOWN - Shutdown Service
-
-Shutdown backend server.
-
-**Request parameters:**
-```python
-body = {}  # No parameters
-```
-
-**Response data:**
-```python
-{
-    "status": "shutdown"
-}
-```
-
---
-
-## Usage Examples
-
-### Basic Usage
-
-```python
-from core import BackendServer, BackendClient
-
-# 1. Create and start backend
-# config_file default: ~/.trulymem/config.json
-server = BackendServer(
-    db_path="graph_memory.db",
-    use_embedded_db=True,
-    config_file=None  # Optional, custom config path
-)
-server.start(
-    api_key="your-api-key",
-    base_url="https://api.deepseek.com",
-    model="deepseek-chat"  # Optional, model name
-)
-
-# 2. Create client
-client = BackendClient(server)
-
-# 3. Send message
-result = client.process_message("Hello")
-if result.get("success"):
-    print(result["content"])
-
-# 4. Shutdown
-client.shutdown()
-```
-
-### Using Packet Protocol
-
-```python
-import queue
-from core import BackendServer, Packet, PacketType
-
-server = BackendServer(config_file=None)
-server.start(api_key="your-key", model="deepseek-chat")
-
-# Create request packet
-response_queue = queue.Queue()
-packet = Packet(
-    id="req-001",
-    type=PacketType.PROCESS_MESSAGE,
-    body={"user_input": "Hello"},
-    response_queue=response_queue
-)
-
-# Send request
-result = server.send(packet)
-print(result.body)
-
-# Shutdown
-server.shutdown()
-```
-
---
-
-## Extension Guide
-
-### Extend to HTTP API
-
-```python
-from flask import Flask, request, jsonify
-from core import BackendServer, BackendClient
-
-app = Flask(__name__)
-server = BackendServer()
-client = BackendClient(server)
-
-@app.route("/message", methods=["POST"])
-def send_message():
-    data = request.json
-    result = client.process_message(data["message"])
-    return jsonify(result)
-
-@app.route("/config", methods=["POST"])
-def update_config():
-    data = request.json
-    result = client.update_settings(
-        api_config=data.get("api_config", {}),
-        tool_limits=data.get("tool_limits", {})
-    )
-    return jsonify(result)
-
-@app.route("/status", methods=["GET"])
-def get_status():
-    result = client.get_status()
-    return jsonify(result)
-
-if __name__ == "__main__":
-    server.start()
-    app.run(port=8080)
-```
-
-### Extend to WebSocket
-
-```python
-import asyncio
-import websockets
-import json
-from core import BackendServer, BackendClient
-
-server = BackendServer()
-client = BackendClient(server)
-
-async def handler(websocket):
-    async for message in websocket:
-        data = json.loads(message)
-        msg_type = data.get("type")
-        
-        if msg_type == "message":
-            result = client.process_message(data["content"])
-        elif msg_type == "settings":
-            result = client.update_settings(
-                api_config=data.get("api_config", {}),
-                tool_limits=data.get("tool_limits", {})
-            )
-        elif msg_type == "status":
-            result = client.get_status()
-        else:
-            result = {"success": False, "error": "unknown type"}
-        
-        await websocket.send(json.dumps(result))
-
-async def main():
-    server.start()
-    async with websockets.serve(handler, "localhost", 8765):
-        await asyncio.Future()
-
-asyncio.run(main())
-```
-
---
-
-## Thread Safety Notes
-
- `BackendServer` uses `threading.Lock` to protect shared resources
- All requests pass through `queue.Queue`, thread-safe
- Responses return through each request's independent response queue
- Default timeout: 30 seconds
-
---
-
-## Tool Call Limits
-
-### Limit Scope
-
-| Call Method | Limited | Description |
-|-------------|---------|-------------|
-| Model-initiated tool calls | ✅ Limited | Triggered via `PROCESS_MESSAGE`, model automatically calls tools |
-| Frontend direct tool calls | ❌ Not limited | Called directly via `EXECUTE_TOOL` |
-
-### Limit Rules (Model-initiated only)
-
-| Category | Operation | Per-Turn Limit |
-|----------|-----------|---------------|
-| Persona graph | Modify | 1 time |
-| Working memory chain | Modify | 5 times |
-| General memory | Query | 20 times |
-| General memory | Modify | 10 times |
-| Context compression | Query | Counted as general memory query |
-
-### Reset Mechanism
-
- Counter resets automatically on each `PROCESS_MESSAGE` call
- Frontend direct `EXECUTE_TOOL` calls do NOT reset the counter
-
---
-
-## Error Handling
-
-All APIs return unified format:
-
-```python
-# Success
-{
-    "success": True,
-    "data": {...}
-}
-
-# Failure
-{
-    "success": False,
-    "error": "Error description"
-}
-```
-
-Common errors:
-
-| Error Message | Description |
-|--------------|-------------|
-| `API Key not configured` | API Key not set |
-| `timeout` | Request timeout |
-| `Tool call rejected: ...` | Tool call rate exceeded limit |
-
-## Web API Endpoints
-
-| Method | Path | Description |
-|--------|------|-------------|
-| GET | /api/check-auth | Check if current session is authenticated |
-| POST | /api/login | Login (JSON body: username, password) |
-| POST | /api/logout | Logout |
-| GET | /api/history | Get chat history |
-| POST | /api/message | Send message to AI |
-| POST | /api/tools/execute | Execute tool call |
-| GET | /api/status | Get system status |
-| GET | /api/settings | Get settings |
-| PUT | /api/settings | Update settings |
-| DELETE | /api/history | Clear history |
-| POST | /api/shutdown | Shutdown server |
-| GET | /api/activity | Get database operation records |
-| GET | /api/graph | Get knowledge graph data |
-| GET | /api/graph/highlight | Get highlighted nodes |
-
-All API endpoints (except /api/login and /api/check-auth) require authentication. Login uses Flask sessions with 7-day validity.
--- a/docs/en/architecture.md
+++ b/docs/en/architecture.md
@ -1,278 +0,0 @@
-# TrulyMEM Architecture
-
-## Core Principles
-
- Keyboard-driven, zero mouse dependency
- Minimalist visual, information density priority
- Tool traces hidden by default, expandable when needed
- TUI & backend separation, multi-threaded communication
- **Everything is a graph**, AI reasoning runs entirely in backend
-
-## Deployment
-
-### Development (Run directly from Git repo)
-
-```bash
-cd TrulyMEM-TrueHumanMEM
-python3 trulymem_entry.py --web --port 4096
-```
-
-### Production (Systemd + standalone directory)
-
-```bash
-# Copy code to standalone deployment directory
-cp -r TrulyMEM-TrueHumanMEM /home/trulymem
-
-# Create Systemd service
-cat > /etc/systemd/system/trulymem-web.service << 'EOF'
-[Unit]
-Description=TrulyMEM - True Human Memory (Web Mode)
-After=network.target
-
-[Service]
-Type=simple
-User=root
-WorkingDirectory=/home/trulymem
-ExecStart=/usr/bin/python3 /home/trulymem/trulymem_entry.py --web --port 4096
-Restart=always
-RestartSec=5
-StandardOutput=journal
-StandardError=journal
-
-[Install]
-WantedBy=multi-user.target
-EOF
-
-systemctl daemon-reload
-systemctl enable trulymem-web.service
-systemctl start trulymem-web.service
-
-# Check status
-systemctl status trulymem-web.service
-```
-
-> **Note**: Do not run the service directly from the Git repository to avoid polluting it with runtime artifacts (logs, databases, etc.).
-
-### Web Access
-
-The service runs at `http://localhost:4096`. On first visit, you'll need to set up an admin account and log in.
-
-### Updating Deployment
-
-```bash
-cd TrulyMEM-TrueHumanMEM
-git pull
-cp -r * /home/trulymem/
-systemctl restart trulymem-web.service
-```
-
---
-
-## Project Structure
-
-```
-TrulyMEM-TrueHumanMEM/
-├── trulymem_entry.py    # Entry: start core → then ui
-├── core/               # Backend/business logic
-│   ├── __init__.py     # Export BackendServer, BackendClient, EmbeddedGraphDB
-│   ├── server.py       # BackendServer (Packet communication protocol)
-│   ├── client.py       # BackendClient (Packet protocol client)
-│   ├── embedded_db.py  # SQLite graph database implementation
-│   ├── graph_client.py # OpenAI/DeepSeek API client
-│   ├── tool_executor.py # Tool executor
-│   ├── tool_limiter.py # Tool call limiter
-│   ├── web_api.py      # Web API service (login + RESTful API)
-│   ├── tools/          # Tool definitions
-│   │   └── memory_tools.py
-│   └── prompts/        # Prompt management (PromptManager + system_prompt.md)
-├── ui/                 # TUI display layer + Web frontend
-│   ├── __init__.py     # Export GraphMemoryApp
-│   ├── app.py          # GraphMemoryApp (communicates via BackendClient)
-│   ├── widgets/        # TUI components
-│   ├── models/         # Data models
-│   ├── services/       # Service layer (config only)
-│   ├── handlers/      # Event handlers
-│   ├── styles/         # Style files
-│   ├── static/         # Web frontend static files
-│   │   ├── graph.html   # Star map visualization (Three.js)
-│   │   └── index.html   # Web chat interface
-│   ├── templates/      # Page templates
-│   │   ├── login.html
-│   │   ├── setup.html
-│   │   └── settings.html
-│   ├── web_config.json         # Web service config file
-│   └── web_config.example.json  # Web config template
-├── tests/              # Test suite
-│   ├── test_core/      # Core logic tests
-│   ├── test_ui/        # UI layer tests
-│   └── test_integration/ # Integration tests
-├── docs/               # Documentation
-│   ├── zh/             # Chinese docs
-│   └── en/             # English docs
-└── build/              # Build scripts
-    ├── build_linux.sh
-    ├── build_macos.sh
-    ├── build_windows.bat
-    ├── build_appimage.sh
-    └── trulymem.spec
-```
-
-## Architecture Diagram
-
-```
-trulymem_entry.py
-    │
-    ├─ BackendServer.start() → Runs in independent thread
-    │   ├─ Handle PROCESS_MESSAGE requests → AI reasoning + tool calls
-    │   ├─ Handle EXECUTE_TOOL requests → External tool calls (unlimited)
-    │   ├─ Handle GET/SET_CONFIG requests
-    │   └─ Manage GraphMemoryClient, EmbeddedGraphDB
-    │
-    └─ GraphMemoryApp(backend_server=server)
-            │
-            └─ BackendClient ← Packet communication → BackendServer
-```
-
-## Component Responsibilities
-
-### core/ (Backend)
-
-| Component | Responsibility |
-|------------|----------------|
-| `server.py` | Packet protocol, multi-threaded queue, AI reasoning, tool limits |
-| `client.py` | Client wrapper, UI-backend communication bridge |
-| `embedded_db.py` | SQLite graph database CRUD |
-| `graph_client.py` | OpenAI/DeepSeek API client |
-| `tool_executor.py` | Tool execution logic |
-| `tool_limiter.py` | Tool call rate limit (AI reasoning only) |
-
-### ui/ (Display Layer)
-
-| Component | Responsibility |
-|------------|----------------|
-| `app.py` | Textual app main class, communicates via BackendClient |
-| `services/` | Config management only, no AI logic |
-
-### Communication Protocol
-
-UI and backend interact via **Packet Communication Protocol**:
-
-```python
-from core import BackendServer, BackendClient, Packet, PacketType
-
-# Backend startup
-server = BackendServer(db_path="graph_memory.db", use_embedded_db=True)
-server.start(api_key="your-key")
-
-# Client communication
-client = BackendClient(server)
-result = client.process_message("hello")  # AI reasoning
-result = client.execute_tool("memory_introspect", {})  # Direct tool call
-```
-
---
-
-## Data Flow
-
-```
-User input → InputBox → on_input_box_send_message
-    ↓
-BackendClient.process_message(user_input)
-    ↓
-Packet (type=PROCESS_MESSAGE) → queue.Queue
-    ↓
-BackendServer (independent thread)
-    <20><><EFBFBD>
-GraphMemoryClient.send_message_with_history()
-    ↓
-OpenAI API / DeepSeek API
-    ↓
-execute_tool() + ToolLimiter (limited during AI reasoning)
-    ↓
-EmbeddedGraphDB (graph database)
-    ↓
-Loop API calls until no tool_calls
-    ↓
-Packet response returns
-    ↓
-MessageHistory displays
-```
-
---
-
-## Startup Flow
-
-```python
-# trulymem_entry.py
-def main():
-    # Config path (~/.trulymem/config.json or project directory)
-    CONFIG_PATH = Path.home() / ".trulymem" / "config.json"
-    DB_PATH = Path.home() / ".trulymem" / "graph_memory.db"
-    
-    # Create backend (config managed by backend)
-    backend_server = BackendServer(
-        db_path=str(DB_PATH),
-        use_embedded_db=True,
-        config_file=str(CONFIG_PATH)
-    )
-    backend_server.start()  # Auto loads config
-    
-    # Create UI (communicates via BackendClient)
-    app = GraphMemoryApp(backend_server=backend_server, config_file=str(CONFIG_PATH))
-    app.run()
-    
-    backend_server.shutdown()
-```
-
---
-
-## Tool System
-
-### Memory Tools (7)
- `memory_recall` - Retrieve memory
- `memory_commit` - Write memory
- `memory_purge` - Delete memory
- `memory_introspect` - View status
- `memory_archive` - Archive memory
- `memory_cleanup` - Clean data
- `context_rewrite` - Compress single-turn tool call context
-
-### Persona Tools (2)
-| `persona_remove` | Delete single persona attribute | Keep other attributes unchanged |
- `persona_update` - Update persona
- `persona_clear` - Clear persona
-
-### Task Tools (6)
-| `task_archive` | Archive completed/expired tasks | Step 6 mandatory, writes completion summary |
-| `task_query` | Query recent task list | Call first in new conversations to avoid duplicate tasks |
- `task_create` - Create task
- `task_set_state` - Set state
- `task_delete` - Delete task
- `task_link_info` - Link information
-
---
-
-## Tool Call Limits
-
-| Category | Operation | Per-Turn Limit |
-|----------|-----------|---------------|
-| Persona graph | Modify | 1 time |
-| Working memory chain | Modify | 5 times |
-| General memory | Query | 20 times |
-| General memory | Modify | 10 times |
-
-> Note: `memory_recall` is uniformly counted as general memory query, no longer distinguished by persona/working memory queries.
-
---
-
-## Error Handling Principle
-
-All APIs **do not throw exceptions**, errors are passed via return dictionary:
-
-```python
-result = client.process_message("hello")
-
-if result.get("success"):
-    print(result["content"])
-else:
-    print(result["error"])  # Error description
--- a/docs/en/memory.md
+++ b/docs/en/memory.md
@ -1,245 +0,0 @@
-# TrulyMEM Memory Mechanism
-
-This document explains the internal memory working mechanism of TrulyMEM.
-
-## Core Design Philosophy
-
-### Different from Traditional Context System
-
-Traditional AI chat systems store conversation history in a messages array:
- Each request carries all historical messages
- Context grows with conversation turns
- Eventually triggers memory compression or sliding window, causing memory loss
-
-TrulyMEM's solution:
- **Abandon** messages array context
- **Only** memory source: Graph database
- All memories stored as triplets (node) - relation → (node)
-
-### Graph Database as the Only Memory Source
-
-All memory must be written to the graph database:
- `memory_commit` - Write new memory
- `memory_purge` - Delete/correct memory
-
-All memory must be read from:
- `memory_recall` - Retrieve memory
-
-### Working Memory Management (Experimental)
-
-`context_rewrite` allows AI to proactively compress tool call context within a single turn:
- Distills verbose JSON tool results into concise natural language summaries
- Summary must include which tools were called and how many calls are summarized
- After system validates the format, replaces `messages_history` with `[user message, summary]`
- Ensures LLM retains meta-cognition (knows "I called tools") while reducing JSON noise
-
---
-
-## Mandatory Execution Flow (Per Turn)
-
-Since there's no traditional context system, each conversation turn must execute in order:
-
-### Step 1: Query Persona Graph (Highest Priority)
-
-```python
-memory_recall(
-    query_intent="AI,persona,role,character,tone,speaking_style",
-    depth=2
-)
-```
-
-**Purpose**: Get current persona, ensure character consistency.
-
-**Processing logic**:
- Persona found → Reply strictly according to persona's tone, style, traits
- Not found → Use default TrulyMEM identity
-
-### Step 2: Query Working Memory Chain
-
-```python
-memory_recall(
-    query_intent="TaskNode,working_memory,task_chain",
-    depth=2
-)
-```
-
-**Purpose**: Get previous task context, understand conversation history.
-
-### Step 3: Process Conversation
-
- Understand user intent
- Generate reply based on persona and working memory chain
- Execute other necessary memory operations
-
-### Step 4: Update Working Memory Chain
-
-```python
-task_create(
-    task_id="Task_current_turn_ID",
-    description="This turn's conversation summary",
-    info_nodes=["related memory nodes"]
-)
-```
-
-**Purpose**: Record this turn's conversation, maintain time chain.
-
---
-
-## Memory Write Rules
-
-### Must-Write Scenarios
-
-The following information **must** be written to the graph database:
-
-| Scenario | Example | Write Method |
-|----------|---------|--------------|
-| User explicitly states preference | "I like rock" | `memory_commit` |
-| User shares information | "I'm working on X project" | `memory_commit` |
-| User makes plans | "I plan to X" | `memory_commit` |
-| User describes state | "I'm currently at X" | `memory_commit` |
-
-### Must-Not-Write Scenarios
-
-The following information **must NOT** be written:
-
-| Scenario | Reason | Handling |
-|----------|--------|----------|
-| AI-inferred user preference | Unverified | Don't write or mark [speculation] |
-| AI-guessed user intent | Unverified | Don't write or mark [speculation] |
-| AI-derived conclusion | Unverified | Don't write or mark [speculation] |
-
-### Annotation Rules
-
-| Type | Annotation | Example |
-|------|------------|---------|
-| Inferred content | Must mark **[speculation]** | user[speculation] likes music |
-| Explicit content | State directly | user likes music |
-
---
-
-## Node & Edge Types
-
-### Node Types
-
-| Node Type | Description | Stores |
-|-----------|-------------|--------|
-| `PersonaNode` | Persona node | AI role, character, tone |
-| `TaskNode` | Task node | Task summary |
-| `StateNode` | State node | Task state |
-| `InfoNode` | Information node | Specific information |
-| `EntityNode` | Entity node | General entity |
-
-### Edge Types
-
-| Edge Type | Description | Relationship |
-|-----------|-------------|--------------|
-| `HAS_PERSONA` | Persona | AI → PersonaNode |
-| `NEXT_TASK` | Time chain | TaskNode → TaskNode |
-| `HAS_STATE` | State | TaskNode → StateNode |
-| `CONTAINS_INFO` | Information | TaskNode → InfoNode |
-| `RELATES_TO` | Related | EntityNode → EntityNode |
-
---
-
-## Must Query Working Memory Chain Scenarios
-
-### Mandatory Query Scenarios
-
-The following scenarios **must** query the working memory chain:
-
-| Scenario | Example |
-|----------|---------|
-| Start of each turn | Execute Step 2 |
-| User mentions "刚才/just now" | "What did we talk about just now?" |
-| User mentions "之前/before" | "Continue the previous topic" |
-| User mentions "上次/last time" | "What we said last time X" |
-| User asks about history | "What did we talk about before?" |
-| Resume continuous task | User returns to previous topic |
-| Context reference | "that thing" |
-
---
-
-## Autonomy Principles
-
-Outside the mandatory execution flow, the AI can freely decide:
-
-### Query Decisions
-
- User asks about history → Query
- Involves previous content → Query
- Uncertain → Can query
-
-### Write Decisions
-
- User explicitly mentions → Must write
- AI infers → Can write, mark [speculation]
-
-### Tool Usage Decisions
-
- Flexible choice based on context
- Avoid over-use
- Maintain natural conversation
-
---
-
-## Complete Example: Idiom Chain Game
-
-### Turn 1: User starts game
-
-```
-User: Let's play idiom chain! I'll start: 为所欲为
-
-AI Actions:
-1. Query persona graph → Get current persona (e.g., catgirl)
-2. Query working memory chain → No active task
-3. Record game state:
-   memory_commit(triplets=[
-     {"subject": "idiom_chain_current", "relation": "content", "object": "为所欲为"},
-     {"subject": "idiom_chain_current", "relation": "game", "object": "idiom_chain"}
-   ])
-4. Create task:
-   task_create(task_id="Task_idiom_chain", description="Idiom chain game, current: 为所欲为", info_nodes=["idiom_chain_current"])
-5. Reply: "Okay~! My turn: 为虎作伥 ~!"
-```
-
-### Turn 2: Topic interrupted
-
-```
-User: Nagato Yuki
-
-AI Actions:
-1. Query persona graph → Get current persona (catgirl)
-2. Query working memory chain → Found Task_idiom_chain status "in_progress"
-3. Pause task:
-   task_set_state(task_id="Task_idiom_chain", state="paused")
-4. Create new task:
-   task_create(task_id="Task_nagato_yuki", description="Discuss Nagato Yuki")
-5. Reply about Nagato Yuki
-```
-
-### Turn 3: User asks to continue game
-
-```
-User: About the idiom chain just now, I don't know how to connect to your idiom, please help me
-
-AI Actions:
-1. Query persona graph → Get current persona (catgirl)
-2. Query working memory chain → Found Task_idiom_chain status "paused"
-3. Resume task:
-   task_set_state(task_id="Task_idiom_chain", state="in_progress")
-4. Query info node → Get current idiom "为虎作伥"
-5. Reply: "Okay~! The last idiom was '为虎作伥', your turn: 伥鬼害人 ~!"
-```
-
---
-
-## Execution Checklist
-
-Must check each conversation turn:
-
- [ ] Step 1: Did you query the persona graph?
- [ ] Step 2: Did you query the working memory chain?
- [ ] Step 3: Did you generate reply based on persona and working memory chain?
- [ ] Step 4: Did you update the working memory chain?
- [ ] Did you query working memory chain when context was referenced?
- [ ] Did you query working memory chain when user mentioned "just now/before/last time"?
--- a/docs/en/persona.md
+++ b/docs/en/persona.md
@ -1,214 +0,0 @@
-# TrulyMEM Persona Graph Mechanism
-
-This document explains the Persona Graph mechanism in TrulyMEM.
-
-## Overview
-
-The Persona Graph is one of TrulyMEM's core mechanisms for maintaining AI's role, character, tone, and other attributes. Different from traditional AI, TrulyMEM's persona is persistent and dynamically switchable, stored in the graph database.
-
-## Core Concepts
-
-### Persona Node (PersonaNode)
-
-Stores AI's role attributes:
-
-| Attribute | Description | Example |
-|-----------|-------------|----------|
-| Role | Current role played | Catgirl, Teacher, Assistant |
-| Speaking Style | Tone characteristics | Cute, Professional, Serious |
-| Personality | Character description | Lively, Strict, Patient |
-| Catchphrase | Habitual phrases | Meow~, Got it |
-| Background | Role background | Catgirl from the stars |
-
-### Persona Edges
-
-| Edge Type | Description | Relationship |
-|----------|-------------|--------------|
-| `HAS_PERSONA` | Persona | AI → PersonaNode |
-
---
-
-## Mandatory Query Mechanism
-
-### Must Execute Per Turn
-
-According to `system_prompt.md`, each conversation turn **must** first query the persona graph:
-
-```python
-memory_recall(
-    query_intent="AI,persona,role,character,tone,speaking_style",
-    depth=2
-)
-```
-
-**Processing logic:**
- Persona found → Reply strictly according to persona's tone, style, traits
- Not found → Use default TrulyMEM identity
-
-### Persona Priority
-
- **Persona priority > default identity**
- Every sentence matches persona's tone, style, traits
- Never break character unless user explicitly asks
-
---
-
-## Tools
-
-### persona_update
-
-Update persona. Modify AI's role, character, tone, etc.
-
-**Parameters:**
-
-| Parameter | Type | Description | Required |
-|-----------|------|-------------|----------|
-| `attributes` | array | Persona attribute list | ✅ |
-| `mode` | string | replace=replace, merge=merge | ❌ |
-
-**attributes sub-parameters:**
-
-| Sub-parameter | Description |
-|---------------|-------------|
-| `attribute` | Attribute name (role, speaking_style, personality, catchphrase, background) |
-| `value` | Attribute value |
-
-**Example - Switch to catgirl role:**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "role", "value": "catgirl"},
-        {"attribute": "speaking_style", "value": "cute, uses 'meow' as filler"},
-        {"attribute": "personality", "value": "lively, clingy, loyal"}
-    ],
-    mode="replace"
-)
-```
-
-**Example - Add new attribute (preserve existing):**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "catchphrase", "value": "meow~"}
-    ],
-    mode="merge"
-)
-```
-
-**Example - Set professional role:**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "role", "value": "Python expert"},
-        {"attribute": "speaking_style", "value": "professional, concise, rich code examples"},
-        {"attribute": "personality", "value": "strict, patient, helpful"}
-    ],
-    mode="replace"
-)
-```
-
-### persona_clear
-
-Clear persona. Delete AI's role settings, restore default identity.
-
-**Parameters:**
-
-| Parameter | Type | Default | Description |
-|-----------|------|---------|-------------|
-| `confirm` | boolean | true | Confirm clear |
-
---
-
-## Update Flow
-
-### When User Requests Role-Playing
-
-1. Use `persona_update` to update persona
-2. Reply immediately according to new persona
-
-### When User Requests Restoring Default
-
-1. Use `persona_clear` to clear persona
-2. Restore to TrulyMEM default identity
-
---
-
-## Conversation Examples
-
-### Example 1: Switch Role
-
-```
-User: Hello, I want you to play a catgirl
-
-AI:
-1. Call persona_update:
-   {
-     "attributes": [
-       {"attribute": "role", "value": "catgirl"},
-       {"attribute": "speaking_style", "value": "cute, uses 'meow' as filler"},
-       {"attribute": "personality", "value": "lively, clingy, loyal"}
-     ],
-     "mode": "replace"
-   }
-2. Call memory_commit to store persona in graph database
-3. Reply: "Okay meow! Hello master~ I'm your catgirl, what do you need help with meow?"
-```
-
-### Example 2: Maintain Role Consistency
-
-```
-User: How's the weather today?
-
-AI: Query persona graph → Get current persona (catgirl)
-Reply: "Meow~ Master, the weather is great today meow! Sunny and perfect for going outside~"
-```
-
-### Example 3: Restore Default
-
-```
-User: Okay, back to normal
-
-AI:
-1. Call persona_clear(confirm=true)
-2. Call memory_purge to delete persona node
-3. Reply: "Okay, restored. I am TrulyMEM, an AI assistant with long-term memory capabilities."
-```
-
---
-
-## Storage Structure
-
-### In Graph Database
-
-```python
-# Persona node
-{
-  "node_type": "PersonaNode",
-  "name": "AI_Persona",
-  "attributes": {
-    "role": "catgirl",
-    "speaking_style": "cute, uses 'meow' as filler",
-    "personality": "lively, clingy, loyal"
-  }
-}
-
-# Edge
-{
-  "edge_type": "HAS_PERSONA",
-  "from": "AI",
-  "to": "AI_Persona"
-}
-```
-
---
-
-## Implementation Points
-
-1. **Mandatory per turn**: Persona graph query is the first step of each conversation
-2. **Persistent storage**: Persona stored in graph database, not lost
-3. **Dynamic switching**: Supports real-time role switching
-4. **Immediate response**: Reply immediately according to new persona after switch
-5. **Clear boundaries**: Never break character unless user explicitly asks
--- a/docs/en/prompts.md
+++ b/docs/en/prompts.md
@ -1,130 +0,0 @@
-# Prompt Manager Documentation
-
-This document describes the prompt management module.
-
-## Overview
-
-The prompt management module (`core/prompts/`) is responsible for loading and managing system prompts that tell the AI how to use memory tools.
-
-## Core Components
-
-| Component | Description |
-|-----------|-------------|
-| `PromptManager` | Prompt manager, singleton pattern |
-| `system_prompt.md` | Main system prompt template |
-
-## Usage
-
-```python
-from core.prompts import PromptManager
-
-# Get singleton instance
-prompt_manager = PromptManager()
-
-# Get system prompt
-system_prompt = prompt_manager.get_system_prompt()
-```
-
-## System Prompt Content
-
-The system prompt contains:
-
-### 1. Core Identity
-
- **Name**: TrulyMEM (TrueHumanMEM)
- **Capability**: Long-term memory based on graph database
- **Philosophy**: Make AI's memory more human-like
-
-### 2. Core Capabilities
-
-1. **Long-term Memory** - Graph database stores entity relationships
-2. **Persona Management** - Role-playing and character settings
-3. **Task Tracking** - Working memory chain
-
-### 3. Memory Principles
-
- **Must write**: User-explicit preferences, shared information, plans
- **Must not write**: AI-inferred content (unless marked [speculation])
- **Annotation**: Inferred content must be marked **[speculation]**
-
-### 4. Mandatory Execution Flow (Per Turn)
-
-```
-Step 1: Query persona graph (highest priority)
-Step 2: Query working memory chain
-Step 3: Process conversation
-Step 4: memory_commit (write key info) → Persist user-explicit important information to the graph database
-Step 5: Update working memory chain
-```
-
-### 5. Tool System
-
-#### Memory Tools
-
-| Tool | Function |
-|------|----------|
-| `memory_recall` | Retrieve memory |
-| `memory_commit` | Write memory |
-| `memory_purge` | Delete memory |
-| `memory_introspect` | View status |
-| `memory_archive` | Archive memory |
-| `memory_cleanup` | Clean data |
-| `context_rewrite` | Compress single-turn tool call context |
-
-#### Persona Tools
-
-| Tool | Function |
-|------|----------|
-| `persona_update` | Update persona |
-| `persona_clear` | Clear persona |
-
-#### Task Tools
-
-| Tool | Function |
-|------|----------|
-| `task_create` | Create task |
-| `task_set_state` | Set state |
-| `task_delete` | Delete task |
-| `task_link_info` | Link information |
-
-### 6. Autonomy Principles
-
-The AI can autonomously decide:
- Whether to query other memories
- Whether to write other memories
- How to use tools (outside mandatory requirements)
-
-### 7. Conversation Style
-
- Natural and smooth
- Avoid mechanical tool calls
- Prioritize understanding user intent
- Use memory to enhance experience when appropriate
-
-## File Structure
-
-```
-core/prompts/
-├── __init__.py           # Export PromptManager
-├── prompt_manager.py      # PromptManager class
-└── templates/
-    └── system_prompt.md # Main system prompt
-```
-
-## Customization
-
-### Customizing System Prompt
-
-Modify `core/prompts/templates/system_prompt.md` to customize the AI's behavior.
-
-### Adding Custom Prompts
-
-1. Add prompt template file to `core/prompts/templates/`
-2. Modify `PromptManager` to support multiple prompts
-3. Use `set_prompt()` to switch prompts
-
-## Caching
-
- System prompts are cached in memory after first load
- `get_system_prompt()` returns cached content
- Cache is per-process, not persisted
--- a/docs/en/quick_start.md
+++ b/docs/en/quick_start.md
@ -1,202 +0,0 @@
-# TrulyMEM Quick Start Guide
-
-> **Version**: Multi-user (v2) — TUI login, user isolation, embedded Web server
-
---
-
-## Running
-
-### Quick Start
-
-```bash
-# From source
-python trulymem_entry.py
-
-# Packaged binary
-./dist/TrulyMEM
-```
-
-### First Run — Login Flow
-
-On first launch, TrulyMEM checks for legacy data and presents a **login screen**:
-
-1. **Clean install** → Enter username/password (first user becomes admin)
-2. **Legacy upgrade** → Detects `~/.trulymem/config.json`, guides migration setup
-3. **Returning user** → Login directly
-
-> 💡 All user data is isolated: `~/.trulymem/{username}/`
-
-### Chat Configuration
-
-After login, press **F2** to open the right-side configuration panel:
-
-1. **API Key** — Required (DeepSeek, OpenAI, etc.)
-2. **Model** — Optional
-3. **Base URL** — Optional
-
-Config saves automatically.
-
---
-
-## Web Visualization
-
-The Web service now runs **embedded in the main process** (no separate subprocess needed).
-
-### Start via TUI (Admin only)
-
-Admin users: press F2 → check "Enable Web Service".
-
-### Start Manually
-
-```bash
-python -m core.web_api --port 4096
-# Visit http://localhost:4096
-```
-
-### First Visit Flow
-
-1. Open `http://localhost:4096` in browser
-2. **No users** → Auto-redirect to setup page, create admin account
-3. **Has users** → Login page
-4. After login → Star map visualization
-
-### Web Features
-
-| Page | Access | Feature |
-|------|--------|---------|
-| 🌟 Star Map | All logged-in | Browse knowledge graph |
-| ⚙ Settings | All logged-in | Change password |
-| 🧑‍💼 User Management | **Admin only** | Add/delete users |
-
---
-
-## Multi-User System
-
-### Directory Layout
-
-```
-~/.trulymem/
-├── trulymem.db          # Global user database (web_users table)
-├── .migrated            # Migration flag
-├── admin/
-│   ├── config.json      # Admin config
-│   └── admin_graph.db   # Admin knowledge graph
-└── user2/
-    ├── config.json      # user2 config
-    └── user2_graph.db   # user2 knowledge graph
-```
-
-### Role Matrix
-
-| Feature | User | Admin |
-|---------|------|-------|
-| Change password | ✅ | ✅ |
-| Configure API Key / Model | ✅ | ✅ |
-| Web service toggle (TUI) | ❌ | ✅ |
-| Web login credentials | ❌ | ✅ |
-| View user list | ❌ | ✅ |
-| Add/delete users | ❌ | ✅ |
-
-> ⚠️ First registered user becomes admin automatically. Add users via Web settings page.
-
---
-
-## Keyboard Shortcuts
-
-| Key | Action |
-|-----|--------|
-| F1 | Help |
-| F2 | Toggle sidebar (config panel) |
-| F3 | Tool details |
-| F5 | Clear screen |
-| F6 | Quit |
-
---
-
-## Building
-
-```bash
-# Linux
-bash build/build_linux.sh
-
-# macOS
-bash build/build_macos.sh
-
-# Windows
-build\build_windows.bat
-
-# AppImage
-bash build/build_appimage.sh
-```
-
-Output: `dist/TrulyMEM` (single binary — TUI and Web server embedded)
-
-> 📦 Since v2, the Web server runs as a thread inside the main process. No need for a separate `trulymem-web` binary.
-
---
-
-## Architecture
-
-### Communication
-
-```
-TUI (Textual)  ←→  BackendClient  ←→  queue.Queue  ←→  BackendServer (thread)
-```
-
-### Config Management
-
- **Per-user**: `~/.trulymem/{username}/config.json`
- **Web config**: `~/.trulymem/trulymem.db` (web_users table)
- **Auto-load**: reads config for logged-in user on startup
- **Persistent**: saves automatically on change
-
-### Web Service Architecture
-
-```
-┌──────────────────────┐
-│   TrulyMEM Process   │
-│  ┌──────┐ ┌────────┐ │
-│  │ TUI  │ │ Flask  │ │  ← Same process, different threads
-│  │      │ │ Thread │ │
-│  └──────┘ └────────┘ │
-└──────────────────────┘
-```
-
---
-
-## FAQ
-
-### Python not found
-
-Install Python 3.8+: https://www.python.org/downloads/
-
-### Dependency installation fails
-
-```bash
-python -m venv venv
-source venv/bin/activate  # Linux/macOS
-venv\Scripts\activate     # Windows
-pip install -r requirements.txt
-```
-
-### Invalid API Key
-
-Check format and whitespace. Reconfigure in TUI sidebar.
-
-### Lost admin account
-
-The first registered user is always admin. If all users lost admin, delete `trulymem.db` from the user directory and re-register.
-
-### Legacy data migration
-
-When old `~/.trulymem/config.json` and `graph_memory.db` are detected, TUI auto-enters migration flow. Legacy files are preserved.
-
---
-
-## Dev Commands
-
-```bash
-pip install -r requirements.txt
-pytest tests/
-bash build/build_linux.sh
-```
--- a/docs/en/working_memory.md
+++ b/docs/en/working_memory.md
@ -1,182 +0,0 @@
-# TrulyMEM Working Memory Chain Mechanism
-
-## Overview
-
-TrulyMEM maintains conversation continuity through the working memory chain mechanism. Since there's no traditional message history array, the graph database is the only memory carrier, making the working memory chain the key mechanism for maintaining conversation context.
-
-## Core Problems
-
-Traditional AI chat systems have these problems when handling continuous tasks:
-
-1. **No working memory chain**: AI cannot remember the current task status being processed
-2. **Task context lost**: When a topic is interrupted, AI cannot recover the previous task
-3. **Lack of task state management**: No clear marking of task completion status
-
-### Problem Example
-
-```
-User: Let's play idiom chain! I'll start with 为所欲为
-AI: Okay! My turn: 为虎作伥!
-
-User: Nagato Yuki  (topic interrupted)
-AI: (discusses Nagato Yuki)
-
-User: About the idiom chain just now, I don't know how to connect to your idiom
-AI: [Guessing] It seems we haven't played an idiom chain game before...
-```
-
-**Problem**: AI completely forgot the previous idiom chain game.
-
-## Solution
-
-### Dedicated Tools
-
-The system provides 4 dedicated task tools:
-
-| Tool | Function | Use Case |
-|------|----------|----------|
-| `task_create` | Create task node | Start new task |
-| `task_set_state` | Set task state | Update in_progress/completed/paused/cancelled |
-| `task_delete` | Delete task | Clean up completed task |
-| `task_link_info` | Link info node | Connect task with specific information |
-
-### Task States
-
- **in_progress**: Task is executing
- **completed**: Task completed successfully
- **paused**: Task interrupted, can be resumed
- **cancelled**: Task cancelled
-
-## Usage Flow
-
-### Must Execute Per Turn
-
-1. **Query persona graph** (highest priority)
-   ```
-   Call memory_recall
-   Parameters: {"query_intent": "AI,persona,role,character,tone", "depth": 2}
-   ```
-
-2. **Query working memory chain**
-   ```
-   Call memory_recall
-   Parameters: {"query_intent": "TaskNode,working_memory,task_chain", "depth": 2}
-   ```
-
-3. **Generate reply based on context**
-
-4. **Update working memory chain** (if necessary)
-
-## Complete Example: Idiom Chain Game
-
-### Turn 1: User starts game
-
-```
-User: Let's play idiom chain! I'll start with 为所欲为
-
-AI Actions:
-1. Query persona graph → Get current persona (e.g., catgirl)
-2. Query working memory chain → No active task
-3. Record game state:
-   Call memory_commit
-   Parameters: {
-     "triplets": [
-       {"subject": "idiom_chain_current", "relation": "content", "object": "为所欲为"},
-       {"subject": "idiom_chain_current", "relation": "game", "object": "idiom_chain"}
-     ]
-   }
-4. Create task node:
-   Call task_create
-   Parameters: {
-     "task_id": "Task_idiom_chain",
-     "description": "Idiom chain game, current idiom: 为所欲为",
-     "info_nodes": ["idiom_chain_current"]
-   }
-5. Reply: "Okay~! My turn: 为虎作伥~!"
-```
-
-### Turn 2: Topic interrupted
-
-```
-User: Nagato Yuki
-
-AI Actions:
-1. Query persona graph → Get current persona (catgirl)
-2. Query working memory chain → Found Task_idiom_chain status "in_progress"
-3. Pause task:
-   Call task_set_state
-   Parameters: {"task_id": "Task_idiom_chain", "state": "paused"}
-4. Create new task:
-   Call task_create
-   Parameters: {"task_id": "Task_nagato_yuki", "description": "Discuss Nagato Yuki"}
-5. Reply about Nagato Yuki
-```
-
-### Turn 3: User asks to continue game
-
-```
-User: About the idiom chain just now, I don't know how to connect to your idiom
-
-AI Actions:
-1. Query persona graph → Get current persona (catgirl)
-2. Query working memory chain → Found Task_idiom_chain status "paused"
-3. Resume task:
-   Call task_set_state
-   Parameters: {"task_id": "Task_idiom_chain", "state": "in_progress"}
-4. Query info node → Get current idiom "为虎作伥"
-5. Reply: "Okay~! The last idiom was '为虎作伥', your turn: 伥鬼害人~!"
-```
-
-## API Reference
-
-### task_create
-
-Create task node to track continuous tasks.
-
-```json
-{
-  "task_id": "Task_idiom_chain",
-  "description": "Task overview",
-  "info_nodes": ["associated info node names"]
-}
-```
-
-### task_set_state
-
-Set task state.
-
-```json
-{
-  "task_id": "Task_idiom_chain",
-  "state": "in_progress"  // in_progress/completed/paused/cancelled
-}
-```
-
-### task_delete
-
-Delete task node.
-
-```json
-{
-  "task_id": "Task_idiom_chain",
-  "delete_info_nodes": true  // whether to delete associated info nodes
-}
-```
-
-### task_link_info
-
-Associate info nodes to task.
-
-```json
-{
-  "task_id": "Task_idiom_chain",
-  "info_node_names": ["idiom_chain_current", "idiom_chain_last"]
-}
-```
-
-## Notes
-
-1. **Persona graph has highest priority**: Must query persona graph first each turn
-2. **Working memory chain is the only context carrier**: No traditional message history
-3. **Task state must be updated timely**: Ensure correct state transitions
-4. **Use dedicated tools**: Prefer task_* tools over memory_commit for task-related operations
--- a/docs/zh/README.md
+++ b/docs/zh/README.md
@ -1,31 +0,0 @@
-# TrulyMEM 文档
-
-欢迎来到 TrulyMEM 项目中文文档。
-
-> [Switch to English version](../en/README.md)
-
-## 文档目录
-
-| 文档 | 内容 |
-|------|------|
-| [architecture.md](architecture.md) | 系统架构和技术设计 |
-| [quick_start.md](quick_start.md) | 完整启动指南与配置说明 |
-| [memory.md](memory.md) | 内部记忆工作机制 |
-| [persona.md](persona.md) | 人设图机制 |
-| [working_memory.md](working_memory.md) | 连续性任务处理机制 |
-| [api.md](api.md) | 后端 API 接口文档（供扩展开发） |
-| [prompts.md](prompts.md) | 提示词管理模块 |
-
-## 项目简介
-
-TrulyMEM (TrueHumanMEM) 是一个让 AI 拥有长期记忆能力的图记忆系统，通过图数据库存储实体关系，让 AI 能够像人类一样记忆、回忆和管理信息。
-
-## 核心特性
-
- **长期记忆存储**: 基于 SQLite 内嵌图数据库，开箱即用
- **人设图机制**: 支持角色扮演和性格设定
- **工作记忆链**: 维持对话连贯性的任务跟踪机制
- **TUI 与后端分离**: 多线程 Queue 通信
- **键盘驱动 TUI**: 无需鼠标，全键盘操作
- **跨平台支持**: Windows / Linux / macOS
- **独立部署**: 支持打包为可执行文件
--- a/docs/zh/api.md
+++ b/docs/zh/api.md
@ -1,535 +0,0 @@
-# BackendServer API 文档
-
-本文档描述后端服务器的 API 接口，供开发者扩展其他连接方式（如网络接口、WebSocket 等）。
-
-## 概述
-
-TrulyMEM 后端采用 **Packet 通信协议**，通过 `queue.Queue` 实现线程安全通信。后端在独立线程中运行，处理来自客户端的请求。
-
-### 核心组件
-
-| 组件 | 说明 |
-|------|------|
-| `BackendServer` | 后端服务器，独立线程运行 |
-| `BackendClient` | 客户端封装，提供便捷方法 |
-| `PacketType` | 请求类型枚举 |
-| `Packet` | 数据包（请求） |
-| `PacketResponse` | 数据包响应 |
-
---
-
-## 请求类型 (PacketType)
-
-```python
-class PacketType(Enum):
-    PROCESS_MESSAGE = "process_message"  # 处理消息
-    EXECUTE_TOOL = "execute_tool"        # 执行工具
-    GET_STATUS = "get_status"            # 获取状态
-    GET_SETTINGS = "get_settings"         # 获取完整配置（api_config + tool_limits）
-    SET_SETTINGS = "set_settings"        # 设置完整配置（api_config + tool_limits）
-    GET_HISTORY = "get_history"           # 获取历史
-    SAVE_HISTORY = "save_history"         # 保存历史
-    SHUTDOWN = "shutdown"                 # 关闭服务
-```
-
---
-
-## 数据包格式
-
-### Packet
-
-```python
-@dataclass
-class Packet:
-    id: str                      # 唯一标识
-    type: PacketType             # 请求类型
-    body: Dict[str, Any]        # 请求参数
-    response_queue: queue.Queue # 响应队列（可选）
-    created_at: float           # 创建时间
-```
-
-### PacketResponse
-
-```python
-@dataclass
-class PacketResponse:
-    id: str                      # 对应的请求ID
-    success: bool                # 是否成功
-    data: Any = None             # 返回数据
-    error: Optional[str] = None  # 错误信息
-```
-
---
-
-## API 接口详情
-
-### 1. PROCESS_MESSAGE - 处理消息
-
-发送用户消息，AI 将处理并返回回复（可能包含工具调用）。
-
-**请求参数：**
-```python
-body = {
-    "user_input": str  # 用户输入的消息
-}
-```
-
-**响应数据：**
-```python
-{
-    "success": True,
-    "content": str,           # AI 回复内容
-    "tool_calls": [           # 工具调用记录
-        {
-            "name": str,      # 工具名称
-            "arguments": dict,# 工具参数
-            "result": str     # 工具执行结果
-        }
-    ],
-    "rejected_tools": [       # 被拒绝的工具调用
-        (str, str)            # (工具名, 拒绝原因)
-    ]
-}
-```
-
-**示例：**
-```python
-from core import BackendServer, BackendClient
-
-server = BackendServer(db_path="graph_memory.db", use_embedded_db=True)
-server.start(api_key="your-api-key")
-
-client = BackendClient(server)
-result = client.process_message("你好，请记住我的名字是小明")
-
-if result.get("success"):
-    # 响应数据在 data 字段中
-    print(result["data"]["content"])
-    # 工具调用: result["data"]["tool_calls"]
-    # 被拒绝的工具: result["data"]["rejected_tools"]
-```
-
---
-
-### 2. EXECUTE_TOOL - 执行工具
-
-直接执行指定的记忆工具。
-
-> **注意**：前端直接调用的工具**不受次数限制**，只有模型发起的工具调用才受限制。
-
-**请求参数：**
-```python
-body = {
-    "tool_name": str,     # 工具名称
-    "arguments": dict    # 工具参数
-}
-```
-
-**响应数据：**
-```python
-{
-    "success": True,
-    "result": str  # 工具执行结果
-}
-```
-
-**示例：**
-```python
-result = client.execute_tool("memory_recall", {"query_intent": "用户信息"})
-```
-
---
-
-### 3. GET_STATUS - 获取状态
-
-获取后端运行状态。
-
-**请求参数：**
-```python
-body = {}  # 无参数
-```
-
-**响应数据：**
-```python
-{
-    "running": bool,              # 后端是否运行中
-    "config": dict,              # 当前配置
-    "graph_initialized": bool,   # 图数据库是否初始化
-    "client_initialized": bool   # API 客户端是否初始化
-}
-```
-
-**示例：**
-```python
-status = client.get_status()
-print(status["data"]["running"])  # True
-```
-
---
-
-### 4. GET_SETTINGS - 获取完整配置
-
-获取当前 API 配置和工具限制（一次获取全部）。
-
-**请求参数：**
-```python
-body = {}  # 无参数
-```
-
-**响应数据：**
-```python
-{
-    "api_config": {
-        "api_key": str,   # API Key
-        "base_url": str,  # API Base URL
-        "model": str     # 模型名称
-    },
-    "tool_limits": {
-        "persona_update_max": int,  # 人设图修改上限
-        "task_update_max": int,      # 工作记忆链修改上限
-        "memory_query_max": int,     # 一般记忆查询上限
-        "memory_update_max": int    # 一般记忆修改上限
-    }
-}
-```
-
-**示例：**
-```python
-result = client.get_settings()
-api_config = result["data"]["api_config"]
-tool_limits = result["data"]["tool_limits"]
-```
-
---
-
-### 5. SET_SETTINGS - 设置完整配置
-
-更新 API 配置和工具限制（一次设置全部）。
-
-**请求参数：**
-```python
-body = {
-    "api_config": {
-        "api_key": str,    # API Key
-        "base_url": str,   # API Base URL (默认: https://api.deepseek.com)
-        "model": str       # 模型名称 (默认: deepseek-chat)
-    },
-    "tool_limits": {
-        "persona_update_max": int,  # 人设图修改上限 (≥1)
-        "task_update_max": int,      # 工作记忆链修改上限 (≥1)
-        "memory_query_max": int,     # 一般记忆查询上限 (≥1)
-        "memory_update_max": int     # 一般记忆修改上限 (≥1)
-    }
-}
-```
-
-**响应数据：**
-```python
-{
-    "status": "settings_updated"
-}
-```
-
-**示例：**
-```python
-result = client.update_settings(
-    api_config={
-        "api_key": "sk-xxxxx",
-        "base_url": "https://api.deepseek.com",
-        "model": "deepseek-chat"
-    },
-    tool_limits={
-        "persona_update_max": 2,
-        "task_update_max": 5,
-        "memory_query_max": 30
-    }
-)
-```
-
---
-
-### 6. GET_HISTORY - 获取消息历史
-
-获取保存的消息历史（从数据库读取，用于UI显示，不参与模型推理）。
-
-**请求参数：**
-```python
-body = {}  # 无参数
-```
-
-**响应数据：**
-```python
-{
-    "history": list  # 消息历史列表 [{"role": "user/assistant", "content": "..."}]
-}
-```
-
-**说明：**
- 消息历史存储在数据库 `chat_records` 表中
- 最多返回最近 500 条记录
- 历史消息仅用于 UI 显示，不参与模型推理
-
---
-
-### 7. SAVE_HISTORY - 保存消息历史
-
-保存消息历史到数据库（每次处理消息后自动保存，用户消息和AI回复分别保存）。
-
-**请求参数：**
-```python
-body = {
-    "messages": list  # 消息列表 [{"role": "...", "content": "..."}]
-}
-```
-
-**响应数据：**
-```python
-{
-    "status": "history_saved"
-}
-```
-
-**说明：**
- 消息自动保存到数据库 `chat_records` 表
- 系统自动限制最多保留 500 条记录，超出后自动删除旧记录
- 每次调用 `PROCESS_MESSAGE` 时，会自动保存用户消息和AI回复
- **清空历史**：通过 `SAVE_HISTORY` 传递空消息列表 `messages=[]` 可清空历史，`client.clear_history()` 方法即基于此实现
-
---
-
-### 8. SHUTDOWN - 关闭服务
-
-关闭后端服务器。
-
-**请求参数：**
-```python
-body = {}  # 无参数
-```
-
-**响应数据：**
-```python
-{
-    "status": "shutdown"
-}
-```
-
---
-
-## 使用示例
-
-### 基础使用
-
-```python
-from core import BackendServer, BackendClient
-
-# 1. 创建并启动后端
-# config_file 默认: ~/.trulymem/config.json
-server = BackendServer(
-    db_path="graph_memory.db",
-    use_embedded_db=True,
-    config_file=None  # 可选，自定义配置路径
-)
-server.start(
-    api_key="your-api-key",
-    base_url="https://api.deepseek.com",
-    model="deepseek-chat"  # 可选，模型名称
-)
-
-# 2. 创建客户端
-client = BackendClient(server)
-
-# 3. 发送消息
-result = client.process_message("你好")
-if result.get("success"):
-    print(result["content"])
-
-# 4. 关闭
-client.shutdown()
-```
-
-### 使用 Packet 协议
-
-```python
-import queue
-from core import BackendServer, Packet, PacketType
-
-server = BackendServer(config_file=None)
-server.start(api_key="your-key", model="deepseek-chat")
-
-# 创建请求包
-response_queue = queue.Queue()
-packet = Packet(
-    id="req-001",
-    type=PacketType.PROCESS_MESSAGE,
-    body={"user_input": "你好"},
-    response_queue=response_queue
-)
-
-# 发送请求
-result = server.send(packet)
-print(result.body)
-
-# 关闭
-server.shutdown()
-```
-
---
-
-## 扩展指南
-
-### 扩展为 HTTP API
-
-```python
-from flask import Flask, request, jsonify
-from core import BackendServer, BackendClient
-
-app = Flask(__name__)
-server = BackendServer()
-client = BackendClient(server)
-
-@app.route("/message", methods=["POST"])
-def send_message():
-    data = request.json
-    result = client.process_message(data["message"])
-    return jsonify(result)
-
-@app.route("/config", methods=["POST"])
-def update_config():
-    data = request.json
-    result = client.update_settings(
-        api_config=data.get("api_config", {}),
-        tool_limits=data.get("tool_limits", {})
-    )
-    return jsonify(result)
-
-@app.route("/status", methods=["GET"])
-def get_status():
-    result = client.get_status()
-    return jsonify(result)
-
-if __name__ == "__main__":
-    server.start()
-    app.run(port=8080)
-```
-
-### 扩展为 WebSocket
-
-```python
-import asyncio
-import websockets
-import json
-from core import BackendServer, BackendClient
-
-server = BackendServer()
-client = BackendClient(server)
-
-async def handler(websocket):
-    async for message in websocket:
-        data = json.loads(message)
-        msg_type = data.get("type")
-        
-        if msg_type == "message":
-            result = client.process_message(data["content"])
-        elif msg_type == "settings":
-            result = client.update_settings(
-                api_config=data.get("api_config", {}),
-                tool_limits=data.get("tool_limits", {})
-            )
-        elif msg_type == "status":
-            result = client.get_status()
-        else:
-            result = {"success": False, "error": "unknown type"}
-        
-        await websocket.send(json.dumps(result))
-
-async def main():
-    server.start()
-    async with websockets.serve(handler, "localhost", 8765):
-        await asyncio.Future()
-
-asyncio.run(main())
-```
-
---
-
-## 线程安全说明
-
- `BackendServer` 使用 `threading.Lock` 保护共享资源
- 所有请求通过 `queue.Queue` 传递，线程安全
- 响应通过每个请求独立的响应队列返回
- 默认超时时间：30 秒
-
---
-
-## 工具调用限制
-
-### 限制范围
-
-| 调用方式 | 是否受限 | 说明 |
-|---------|---------|------|
-| 模型发起的工具调用 | ✅ 受限 | 通过 `PROCESS_MESSAGE` 触发，模型自动调用工具 |
-| 前端直接调用工具 | ❌ 不受限 | 通过 `EXECUTE_TOOL` 直接调用 |
-
-### 限制规则（仅限模型发起）
-
-| 类别 | 操作 | 每轮上限 |
-|------|------|---------|
-| 人设图 | 修改 | 1 次 |
-| 工作记忆链 | 修改 | 5 次 |
-| 一般记忆 | 查询 | 20 次 |
-| 一般记忆 | 修改 | 10 次 |
-| 上下文压缩 | 查询 | 计入一般记忆查询 |
-
-### 重置机制
-
- 每次调用 `PROCESS_MESSAGE` 时，计数器自动重置
- 前端直接调用 `EXECUTE_TOOL` 不会重置计数器
-
---
-
-## 错误处理
-
-所有 API 返回统一格式：
-
-```python
-# 成功
-{
-    "success": True,
-    "data": {...}
-}
-
-# 失败
-{
-    "success": False,
-    "error": "错误描述"
-}
-```
-
-常见错误：
-
-| 错误信息 | 说明 |
-|---------|------|
-| `API Key 未配置` | 未设置 API Key |
-| `timeout` | 请求超时 |
-| `工具调用被拒绝: ...` | 工具调用频率超限 |
-
-## Web API 端点
-
-| 方法 | 路径 | 说明 |
-|------|------|------|
-| GET | /api/check-auth | 检查当前会话是否已登录 |
-| POST | /api/login | 登录（JSON body: username, password） |
-| POST | /api/logout | 登出 |
-| GET | /api/history | 获取聊天历史 |
-| POST | /api/message | 发送消息给 AI |
-| POST | /api/tools/execute | 执行工具调用 |
-| GET | /api/status | 获取系统状态 |
-| GET | /api/settings | 获取设置 |
-| PUT | /api/settings | 更新设置 |
-| DELETE | /api/history | 清空历史 |
-| POST | /api/shutdown | 关闭服务器 |
-| GET | /api/activity | 获取数据库操作记录 |
-| GET | /api/graph | 获取知识图谱数据 |
-| GET | /api/graph/highlight | 获取高亮节点 |
-
-所有 API 端点（除 /api/login 和 /api/check-auth 外）需要登录认证。登录使用 Flask session，有效期 7 天。
--- a/docs/zh/architecture.md
+++ b/docs/zh/architecture.md
@ -1,286 +0,0 @@
-# TrulyMEM 架构设计
-
-## 核心原则
-
- 键盘驱动，零鼠标依赖
- 极简视觉，信息密度优先
- 工具痕迹默认隐藏，需要时可展开
- TUI 与后端分离，多线程通信
- **一切皆图**，AI 推理全部在后端
-
-## 部署方式
-
-### 开发环境（从 Git 仓库直接运行）
-
-```bash
-cd TrulyMEM-TrueHumanMEM
-python3 trulymem_entry.py --web --port 4096
-```
-
-### 生产环境（Systemd + 独立部署目录）
-
-```bash
-# 将代码复制到独立目录
-cp -r TrulyMEM-TrueHumanMEM /home/trulymem
-
-# 创建 Systemd 服务
-cat > /etc/systemd/system/trulymem-web.service << 'EOF'
-[Unit]
-Description=TrulyMEM - True Human Memory (Web Mode)
-After=network.target
-
-[Service]
-Type=simple
-User=root
-WorkingDirectory=/home/trulymem
-ExecStart=/usr/bin/python3 /home/trulymem/trulymem_entry.py --web --port 4096
-Restart=always
-RestartSec=5
-StandardOutput=journal
-StandardError=journal
-
-[Install]
-WantedBy=multi-user.target
-EOF
-
-systemctl daemon-reload
-systemctl enable trulymem-web.service
-systemctl start trulymem-web.service
-
-# 查看状态
-systemctl status trulymem-web.service
-```
-
-> **注意**: 不要直接从 Git 仓库启动服务，以免日志文件、数据库等运行时产物污染仓库。
-
-### Web 访问
-
-服务默认运行在 `http://localhost:4096`，首次访问需设置管理员账号并登录。
-
-### 更新部署
-
-```bash
-# 拉取最新代码
-cd TrulyMEM-TrueHumanMEM
-git pull
-
-# 同步到部署目录
-cp -r * /home/trulymem/
-
-# 重启服务
-systemctl restart trulymem-web.service
-```
-
---
-
-## 原始架构说明
-## 项目结构
-
-```
-TrulyMEM-TrueHumanMEM/
-├── trulymem_entry.py    # 入口：先启动 core → 再启动 ui
-├── core/               # 后端/业务逻辑
-│   ├── __init__.py     # 导出 BackendServer, BackendClient, EmbeddedGraphDB
-│   ├── server.py       # BackendServer (Packet 通信协议)
-│   ├── client.py       # BackendClient (Packet 协议客户端)
-│   ├── embedded_db.py  # SQLite 图数据库实现
-│   ├── graph_client.py # OpenAI/DeepSeek API 客户端
-│   ├── tool_executor.py # 工具执行器
-│   ├── tool_limiter.py # 工具调用限制器
-│   ├── web_api.py      # Web API 服务（登录 + RESTful API）
-│   ├── tools/          # 工具定义
-│   │   └── memory_tools.py
-│   └── prompts/        # 提示词管理（PromptManager + system_prompt.md）
-├── ui/                 # TUI 显示层 + Web 前端
-│   ├── __init__.py     # 导出 GraphMemoryApp
-│   ├── app.py          # GraphMemoryApp (通过 BackendClient 通信)
-│   ├── widgets/        # TUI 组件
-│   ├── models/         # 数据模型
-│   ├── services/       # 服务层（仅配置管理）
-│   ├── handlers/      # 事件处理
-│   ├── styles/         # 样式文件
-│   ├── static/         # Web 前端静态文件
-│   │   ├── graph.html   # 星图可视化（Three.js）
-│   │   └── index.html   # Web 聊天界面
-│   ├── templates/      # 页面模板
-│   │   ├── login.html
-│   │   ├── setup.html
-│   │   └── settings.html
-│   ├── web_config.json         # Web 服务配置文件
-│   └── web_config.example.json  # Web 配置模板
-├── tests/              # 测试套件
-│   ├── test_core/      # 核心逻辑测试
-│   ├── test_ui/        # UI 层测试
-│   └── test_integration/ # 集成测试
-├── docs/               # 文档
-│   ├── zh/             # 中文文档
-│   └── en/             # 英文文档
-└── build/              # 打包脚本
-    ├── build_linux.sh
-    ├── build_macos.sh
-    ├── build_windows.bat
-    ├── build_appimage.sh
-    └── trulymem.spec
-```
-
-## 架构图
-
-```
-trulymem_entry.py
-    │
-    ├─ BackendServer.start() → 独立线程运行
-    │   ├─ 处理 PROCESS_MESSAGE 请求 → AI 推理 + 工具调用
-    │   ├─ 处理 EXECUTE_TOOL 请求 → 外部工具调用（不限次数）
-    │   ├─ 处理 GET/SET_CONFIG 请求
-    │   └─ 管理 GraphMemoryClient, EmbeddedGraphDB
-    │
-    └─ GraphMemoryApp(backend_server=server)
-            │
-            └─ BackendClient ← Packet 通信 → BackendServer
-```
-
-## 组件职责
-
-### core/ (后端)
-
-| 组件 | 职责 |
-|------|------|
-| `server.py` | Packet 协议处理，多线程队列通信，AI 推理，工具限制 |
-| `client.py` | 客户端封装，UI 与后端通信桥梁 |
-| `embedded_db.py` | SQLite 图数据库 CRUD |
-| `graph_client.py` | OpenAI/DeepSeek API 客户端 |
-| `tool_executor.py` | 工具执行逻辑 |
-| `tool_limiter.py` | 工具调用频率限制（仅限 AI 推理） |
-
-### ui/ (显示层)
-
-| 组件 | 职责 |
-|------|------|
-| `app.py` | Textual 应用主类，仅通过 BackendClient 通信 |
-| `services/` | 仅配置管理，无 AI 逻辑 |
-
-### 通信协议
-
-UI 与后端通过 **Packet 通信协议** 交互：
-
-```python
-from core import BackendServer, BackendClient, Packet, PacketType
-
-# 后端启动
-server = BackendServer(db_path="graph_memory.db", use_embedded_db=True)
-server.start(api_key="your-key")
-
-# 客户端通信
-client = BackendClient(server)
-result = client.process_message("你好")  # AI 推理
-result = client.execute_tool("memory_introspect", {})  # 外部工具调用
-```
-
---
-
-## 数据流
-
-```
-用户输入 → InputBox → on_input_box_send_message
-    ↓
-BackendClient.process_message(user_input)
-    ↓
-Packet (type=PROCESS_MESSAGE) → queue.Queue
-    ↓
-BackendServer (独立线程)
-    ↓
-GraphMemoryClient.send_message_with_history()
-    ↓
-OpenAI API / DeepSeek API
-    ↓
-execute_tool() + ToolLimiter (AI 推理时受限)
-    ↓
-EmbeddedGraphDB (图数据库)
-    ↓
-循环调用 API 直到无 tool_calls
-    ↓
-Packet 响应返回
-    ↓
-MessageHistory 显示
-```
-
---
-
-## 启动流程
-
-```python
-# trulymem_entry.py
-def main():
-    # 配置文件路径 (~/.trulymem/config.json 或项目目录)
-    CONFIG_PATH = Path.home() / ".trulymem" / "config.json"
-    DB_PATH = Path.home() / ".trulymem" / "graph_memory.db"
-    
-    # 创建后端（配置由后端管理）
-    backend_server = BackendServer(
-        db_path=str(DB_PATH),
-        use_embedded_db=True,
-        config_file=str(CONFIG_PATH)
-    )
-    backend_server.start()  # 自动加载配置
-    
-    # 创建UI（通过 BackendClient 通信）
-    app = GraphMemoryApp(backend_server=backend_server, config_file=str(CONFIG_PATH))
-    app.run()
-    
-    backend_server.shutdown()
-```
-
---
-
-## 工具系统
-
-### 记忆工具 (7个)
- `memory_recall` - 检索记忆
- `memory_commit` - 写入记忆
- `memory_purge` - 删除记忆
- `memory_introspect` - 查看状态
- `memory_archive` - 归档记忆
- `memory_cleanup` - 清理数据
- `context_rewrite` - 压缩单轮工具调用上下文
-
-### 人设工具 (2个)
-| `persona_remove` | 删除单条人设属性 | 保留其他人设不变 |
- `persona_update` - 更新人设
- `persona_clear` - 清除人设
-
-### 任务工具 (6个)
-| `task_archive` | 归档已完成/过期的任务 | 步骤6强制执行，写入完成摘要 |
-| `task_query` | 查询最近任务列表 | 新对话时优先调用，避免重复创建任务 |
- `task_create` - 创建任务
- `task_set_state` - 设置状态
- `task_delete` - 删除任务
- `task_link_info` - 关联信息
-
---
-
-## 工具调用限制
-
-| 类别 | 操作 | 每轮上限 |
-|------|------|---------|
-| 人设图 | 修改 | 1 次 |
-| 工作记忆链 | 查询 | 30 次 |
-| 工作记忆链 | 修改 | 20 次 |
-| 一般记忆 | 查询 | 30 次 |
-| 一般记忆 | 修改 | 15 次 |
-
-> 注：`memory_recall` 统一计入一般记忆查询，不再区分人设/工作记忆查询。
-
---
-
-## 错误处理原则
-
-所有 API **不抛出异常**，错误通过返回字典传递：
-
-```python
-result = client.process_message("hello")
-
-if result.get("success"):
-    print(result["content"])
-else:
-    print(result["error"])  # 错误描述
-```
--- a/docs/zh/memory.md
+++ b/docs/zh/memory.md
@ -1,245 +0,0 @@
-# TrulyMEM 记忆机制
-
-本文档详细说明 TrulyMEM 内部的记忆工作机制。
-
-## 核心设计理念
-
-### 区别于传统上下文系统
-
-传统 AI 对话系统使用 messages 数组存储对话历史：
- 每次请求携带全部历史消息
- 随着对话轮次增加，上下文逐渐膨胀
- 最终触发记忆压缩或滑动窗口，造成记忆丢失
-
-TrulyMEM 的解决思路：
- **摒弃** messages 数组上下文
- **唯一** 记忆载体：图数据库
- 全部记忆以三元组（节点）- 关系 → （节点）形式存储
-
-### 图数据库作为唯一记忆源
-
-所有记忆必须通过以下方式写入图数据库：
- `memory_commit` - 写入新记忆
- `memory_purge` - 删除/修正记忆
-
-所有记忆必须通过以下方式读取：
- `memory_recall` - 检索记忆
-
-### 工作记忆管理
-
-`context_rewrite` 允许 AI 在单轮对话内主动压缩工具调用的临时上下文：
- 将冗长的 JSON 工具结果提炼为简洁的自然语言摘要
- 摘要必须包含调用了哪些工具、对几次调用的总结
- 系统验证格式后，替换 `messages_history` 为 `[用户消息, 摘要]`
- 确保 LLM 保留元认知（知道"我调用过工具"），同时减少 JSON 噪音
-
---
-
-## 强制执行流程（每轮对话）
-
-由于没有传统上下文系统，每轮对话必须按以下顺序执行：
-
-### 步骤 1：查询人设图（最高优先级）
-
-```python
-memory_recall(
-    query_intent="AI,人设,角色,性格,语气,说话风格",
-    depth=2
-)
-```
-
-**目的**：获取当前人设，确保角色一致性。
-
-**处理逻辑**：
- 找到人设 → 严格按照人设的语气、风格、特征回复
- 未找到 → 使用默认 TrulyMEM 身份
-
-### 步骤 2：查询工作记忆链
-
-```python
-memory_recall(
-    query_intent="TaskNode,工作记忆,任务链",
-    depth=2
-)
-```
-
-**目的**：获取之前的任务上下文，了解对话历史。
-
-### 步骤 3：处理对话
-
- 理解用户意图
- 根据人设和工作记忆链生成回复
- 执行其他必要的记忆操作
-
-### 步骤 4：更新工作记忆链
-
-```python
-task_create(
-    task_id="Task_当前轮次ID",
-    description="本轮对话概述",
-    info_nodes=["相关记忆节点"]
-)
-```
-
-**目的**：记录本轮对话，维持时间链。
-
---
-
-## 记忆写入规则
-
-### 必须写入的情况
-
-以下信息**必须**写入图数据库：
-
-| 场景 | 示例 | 写入方式 |
-|------|------|----------|
-| 用户明确偏好 | "我喜欢摇滚" | `memory_commit` |
-| 用户分享信息 | "我在做X项目" | `memory_commit` |
-| 用户制定计划 | "我打算X" | `memory_commit` |
-| 用户描述状态 | "我现在在X" | `memory_commit` |
-
-### 禁止写入的情况
-
-以下信息**禁止**写入：
-
-| 场景 | 原因 | 处理方式 |
-|------|------|----------|
-| AI 推断的用户偏好 | 未经证实 | 不写入或标注[推测] |
-| AI 猜测的用户意图 | 未经证实 | 不写入或标注[推测] |
-| AI 推导的结论 | 未经证实 | 不写入或标注[推测] |
-
-### 标注规则
-
-| 类型 | 标注方式 | 示例 |
-|------|----------|------|
-| 推理内容 | 必须标注 **[猜测]** | 用户[推测]喜欢音乐 |
-| 明确内容 | 直接陈述 | 用户喜欢音乐 |
-
---
-
-## 节点与边类型
-
-### 节点类型
-
-| 节点类型 | 说明 | 存储内容 |
-|----------|------|----------|
-| `PersonaNode` | 人设节点 | AI 角色、性格、语气 |
-| `TaskNode` | 任务节点 | 任务概述 |
-| `StateNode` | 状态节点 | 任务状态 |
-| `InfoNode` | 信息节点 | 具体信息 |
-| `EntityNode` | 实体节点 | 通用实体 |
-
-### 边类型
-
-| 边类型 | 说明 | 连接关系 |
-|----------|------|----------|
-| `HAS_PERSONA` | 人设 | AI → PersonaNode |
-| `NEXT_TASK` | 时间链 | TaskNode → TaskNode |
-| `HAS_STATE` | 状态 | TaskNode → StateNode |
-| `CONTAINS_INFO` | 信息 | TaskNode → InfoNode |
-| `RELATES_TO` | 关联 | EntityNode → EntityNode |
-
---
-
-## 必须查询工作记忆链的场景
-
-### 强制查询场景
-
-以下情况**必须**查询工作记忆链：
-
-| 场景 | 示例 |
-|------|------|
-| 每轮对话开始 | 执行步骤 2 |
-| 用户提到"刚才" | "刚才我们聊了什么？" |
-| 用户提到"之前" | "继续刚才的话题" |
-| 用户提到"上次" | "上次说的X" |
-| 用户询问历史 | "我们之前说了什么？" |
-| 连续性任务恢复 | 用户回到之前的话题 |
-| 上下文引用 | "那个东西" |
-
---
-
-## 自主性原则
-
-在强制执行流程之外，AI 可自主决定：
-
-### 查询决策
-
- 用户询问历史 → 查询
- 涉及之前内容 → 查询
- 不确定时 → 可查询
-
-### 写入决策
-
- 用户明确提到 → 必须写入
- AI 推理得到 → 可以写入，标注[推测]
-
-### 工具使用决策
-
- 根据上下文灵活选择
- 避免过度使用
- 保持自然对话
-
---
-
-## 完整示例：成语接龙游戏
-
-### 第一轮：用户发起游戏
-
-```
-用户: 咱来玩成语接龙吧，我先开始，为所欲为
-
-AI操作:
-1. 查询人设图 → 获取当前人设（如：猫娘）
-2. 查询工作记忆链 → 无进行中任务
-3. 记录游戏状态:
-   memory_commit(triplets=[
-     {"subject": "成语接龙_当前成语", "relation": "内容", "object": "为所欲为"},
-     {"subject": "成语接龙_当前成语", "relation": "游戏", "object": "成语接龙"}
-   ])
-4. 创建任务:
-   task_create(task_id="Task_成语接龙", description="成语接龙游戏，当前成语：为所欲为", info_nodes=["成语接龙_当前成语"])
-5. 回复: "好的喵！我接：为虎作伥喵！"
-```
-
-### 第二轮：话题被打断
-
-```
-用户: 长门有希
-
-AI操作:
-1. 查询人设图 → 获取当前人设（猫娘）
-2. 查询工作记忆链 → 发现 Task_成语接龙 状态为"进行中"
-3. 暂停任务:
-   task_set_state(task_id="Task_成语接龙", state="已暂停")
-4. 创建新任务:
-   task_create(task_id="Task_长门有希", description="讨论长门有希")
-5. 回复关于长门有希的内容
-```
-
-### 第三轮：用户要求继续游戏
-
-```
-用户: 关于刚才的成语接龙，我并不知道应该怎么接你的成语，请帮我接一下
-
-AI操作:
-1. 查询人设图 → 获取当前人设（猫娘）
-2. 查询工作记忆链 → 发现 Task_成语接龙 状态为"已暂停"
-3. 恢复任务:
-   task_set_state(task_id="Task_成语接龙", state="进行中")
-4. 查询信息节点 → 获取当前成语"为虎作伥"
-5. 回复: "好的喵！上一个成语是'为虎作伥'，我帮你接：伥鬼害人喵！"
-```
-
---
-
-## 执行检查清单
-
-每轮对话必须检查：
-
- [ ] 步骤 1：是否查询了人设图？
- [ ] 步骤 2：是否查询了工作记忆链？
- [ ] 步骤 3：是否根据人设和工作记忆链生成回复？
- [ ] 步骤 4：是否更新了工作记忆链？
- [ ] 涉及上下文引用时是否查询了工作记忆链？
- [ ] 用户提到"刚才/之前/上次"时是否查询了工作记忆链？
--- a/docs/zh/persona.md
+++ b/docs/zh/persona.md
@ -1,214 +0,0 @@
-# TrulyMEM 人设图机制
-
-本文档详细说明 TrulyMEM 的人设图（Persona Graph）工作机制。
-
-## 概述
-
-人设图是 TrulyMEM 的核心机制之一，用于维护 AI 的角色、性格、语气等属性。与传统 AI 不同，TrulyMEM 的人设是可持久化、可动态切换的，存储在图数据库中。
-
-## 核心概念
-
-### 人设节点（PersonaNode）
-
-存储 AI 的角色属性：
-
-| 属性 | 说明 | 示例 |
-|------|------|------|
-| 扮演角色 | AI 当前扮演的角色 | 猫娘、教师、助手 |
-| 说话风格 | 语气特点 |可爱、严肃、专业 |
-| 性格特点 | 性格描述 | 活泼、严谨、耐心 |
-| 口头禅 | 习惯用语 | 喵呜~、明白了 |
-| 背景故事 | 角色背景设定 | 来自星海的猫娘 |
-
-### 人设边（Edge）
-
-| 边类型 | 说明 | 连接关系 |
-|------|------|----------|
-| `HAS_PERSONA` | 人设 | AI → PersonaNode |
-
---
-
-## 强制查询机制
-
-### 每轮对话必须执行
-
-根据 `system_prompt.md`，每轮对话**必须**首先查询人设图：
-
-```python
-memory_recall(
-    query_intent="AI,人设,角色,性格,语气,说话风格",
-    depth=2
-)
-```
-
-**处理逻辑：**
- 找到人设 → 严格按照人设的语气、风格、特征回复
- 未找到 → 使用默认 TrulyMEM 身份
-
-### 人设优先级
-
- **人设优先级 > 默认身份**
- 每句话都符合人设的语气、风格、特征
- 绝不主动跳出角色，除非用户明确要求
-
---
-
-## 工具
-
-### persona_update
-
-更新人设。修改 AI 的角色、性格、语气等属性。
-
-**参数：**
-
-| 参数 | 类型 | 说明 | 必填 |
-|------|------|------|------|
-| `attributes` | array | 人设属性列表 | ✅ |
-| `mode` | string | replace=替换, merge=合并 | ❌ |
-
-**attributes 子参数：**
-
-| 子参数 | 说明 |
-|------|------|
-| `attribute` | 属性名（扮演角色、说话风格、性格特点、口头禅、背景故事） |
-| `value` | 属性值 |
-
-**示例 - 切换为猫娘角色：**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "扮演角色", "value": "猫娘"},
-        {"attribute": "说话风格", "value": "可爱、卖萌、使用'喵'作为语气词"},
-        {"attribute": "性格特点", "value": "活泼、粘人、忠诚"}
-    ],
-    mode="replace"
-)
-```
-
-**示例 - 添加新属性（保留现有属性）：**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "口头禅", "value": "喵呜~"}
-    ],
-    mode="merge"
-)
-```
-
-**示例 - 设置专业角色：**
-
-```python
-persona_update(
-    attributes=[
-        {"attribute": "扮演角色", "value": "Python专家"},
-        {"attribute": "说话风格", "value": "专业、简洁、代码示例丰富"},
-        {"attribute": "性格特点", "value": "严谨、耐心、乐于助人"}
-    ],
-    mode="replace"
-)
-```
-
-### persona_clear
-
-清除人设。删除 AI 的角色设定，恢复默认身份。
-
-**参数：**
-
-| 参数 | 类型 | 默认值 | 说明 |
-|------|------|--------|------|
-| `confirm` | boolean | true | 确认清除 |
-
---
-
-## 更新流程
-
-### 用户要求角色扮演时
-
-1. 使用 `persona_update` 更新人设
-2. 立即按照新人设回复
-
-### 用户要求恢复默认身份时
-
-1. 使用 `persona_clear` 清除人设
-2. 恢复为 TrulyMEM 默认身份
-
---
-
-## 对话示例
-
-### 示例 1：切换角色
-
-```
-用户: 你好，我想让你扮演一只猫娘
-
-AI:
-1. 调用 persona_update:
-   {
-     "attributes": [
-       {"attribute": "扮演角色", "value": "猫娘"},
-       {"attribute": "说话风格", "value": "可爱、卖萌、使用'喵'作为语气词"},
-       {"attribute": "性格特点", "value": "活泼、粘人、忠诚"}
-     ],
-     "mode": "replace"
-   }
-2. 调用 memory_commit 存储人设到图数据库
-3. 回复: "好的喵！主人你好呀~我是主人的猫娘，有什么需要帮忙的吗喵？"
-```
-
-### 示例 2：保持角色一致性
-
-```
-用户: 今天的天气怎么样？
-
-AI: 查询人设图 → 获取当前人设（猫娘）
-回复: "喵~主人，今天天气很好呢喵！阳光明媚，很适合出门散步哦~"
-```
-
-### 示例 3：恢复默认身份
-
-```
-用户: 好了，恢复正常吧
-
-AI:
-1. 调用 persona_clear(confirm=true)
-2. 调用 memory_purge 删除人设节点
-3. 回复: "好的，已恢复正常。我是 TrulyMEM，一个拥有长期记忆能力的 AI 助手。"
-```
-
---
-
-## 存储结构
-
-### 图数据库中的存储
-
-```python
-# 人设节点
-{
-  "node_type": "PersonaNode",
-  "name": "AI人设",
-  "attributes": {
-    "扮演角色": "猫娘",
-    "说话风格": "可爱、卖萌、使用'喵'作为语气词",
-    "性格特点": "活泼、粘人、忠诚"
-  }
-}
-
-# 边
-{
-  "edge_type": "HAS_PERSONA",
-  "from": "AI",
-  "to": "AI人设"
-}
-```
-
---
-
-## 实现要点
-
-1. **每轮强制查询**：人设图查询是每轮对话的第一步
-2. **持久化存储**：人设存储在图数据库中，不丢失
-3. **动态切换**：支持实时切换角色
-4. **状态保持**：切换后立即按新人设回复
-5. **明确边界**：除非用户要求，绝不主动跳出角色
--- a/docs/zh/prompts.md
+++ b/docs/zh/prompts.md
@ -1,133 +0,0 @@
-# 提示词管理文档
-
-本文档描述提示词管理模块。
-
-## 概述
-
-提示词管理模块（`core/prompts/`）负责加载和管理告诉 AI 如何使用记忆工具的系统提示词。
-
-## 核心组件
-
-| 组件 | 说明 |
-|------|------|
-| `PromptManager` | 提示词管理器，单例模式 |
-| `system_prompt.md` | 主要系统提示词模板 |
-
-## 使用方法
-
-```python
-from core.prompts import PromptManager
-
-# 获取单例实例
-prompt_manager = PromptManager()
-
-# 获取系统提示词
-system_prompt = prompt_manager.get_system_prompt()
-```
-
-## 系统提示词内容
-
-系统提示词包含：
-
-### 1. 核心身份
-
- **名称**: TrulyMEM (TrueHumanMEM)
- **能力**: 基于图数据库的长期记忆
- **理念**: 让 AI 的记忆方式更像人类
-
-### 2. 核心能力
-
-1. **长期记忆** - 图数据库存储实体关系
-2. **人设管理** - 角色扮演和性格设定
-3. **任务跟踪** - 工作记忆链
-
-### 3. 记忆原则
-
- **必须写入**: 用户明确表达的偏好、分享的信息、计划
- **禁止写入**: AI 推断的内容（除非标注[推测]）
- **标注**: 推断内容必须标注 **[推测]**
-
-### 4. 强制执行流程（每轮）
-
-```
-步骤 1: 查询人设图（最高优先级）
-步骤 2: 查询工作记忆链
-步骤 3: 处理对话
-步骤 4: memory_commit (写入关键信息) → 将用户明确提到的重要信息写入图数据库
-步骤 5: 更新工作记忆链
-```
-
-> **注意**: 原提示词仅包含 4 步，缺少显式的写入步骤，导致 AI 只查不写、聊完即忘。
-> 步骤 4 确保每轮对话的关键信息被持久化到图数据库中。
-
-### 5. 工具系统
-
-#### 记忆工具
-
-| 工具 | 功能 |
-|------|------|
-| `memory_recall` | 检索记忆 |
-| `memory_commit` | 写入记忆 |
-| `memory_purge` | 删除记忆 |
-| `memory_introspect` | 查看状态 |
-| `memory_archive` | 归档记忆 |
-| `memory_cleanup` | 清理数据 |
-| `context_rewrite` | 压缩单轮工具调用上下文 |
-
-#### 人设工具
-
-| 工具 | 功能 |
-|------|------|
-| `persona_update` | 更新人设 |
-| `persona_clear` | 清除人设 |
-
-#### 任务工具
-
-| 工具 | 功能 |
-|------|------|
-| `task_create` | 创建任务 |
-| `task_set_state` | 设置状态 |
-| `task_delete` | 删除任务 |
-| `task_link_info` | 关联信息 |
-
-### 6. 自主性原则
-
-AI 可自主决定：
- 是否查询其他记忆
- 是否写入其他记忆
- 如何使用工具（强制要求外）
-
-### 7. 对话风格
-
- 自然流畅
- 避免机械式工具调用
- 优先理解用户意图
- 适时使用记忆增强体验
-
-## 文件结构
-
-```
-core/prompts/
-├── __init__.py           # 导出 PromptManager
-├── prompt_manager.py     # PromptManager 类
-└── templates/
-    └── system_prompt.md # 主要系统提示词
-```
-
-## 自定义
-
-### 自定义系统提示词
-
-修改 `core/prompts/templates/system_prompt.md` 自定义 AI 行为。
-
-### 添加自定义提示词
-
-1. 在 `core/prompts/templates/` 添加提示词模板文件
-2. 修改 `PromptManager` 支持多个提示词
-3. 使用 `set_prompt()` 切换提示词
-
-## 缓存
-
- 系统提示词首次加载后缓存在内存中
- `get_system_prompt()` 返回缓存内容
- 缓存按进程，不持久化
--- a/docs/zh/quick_start.md
+++ b/docs/zh/quick_start.md
@ -1,200 +0,0 @@
-# TrulyMEM 启动指南
-
-> **版本**: 多用户版 (v2) — 支持 TUI 登录、多用户隔离、Web 服务内嵌
-
---
-
-## 运行方式
-
-### 快速启动（推荐）
-
-```bash
-# 从源码
-python trulymem_entry.py
-
-# 或打包后
-./dist/TrulyMEM
-```
-
-### 首次使用 —— 登录流程
-
-首次启动会自动检查是否需要迁移旧数据，然后进入**登录页面**：
-
-1. **新部署** → 直接输入用户名和密码创建账户（首个用户自动成为管理员）
-2. **旧版升级** → 自动检测 `~/.trulymem/config.json`，引导设置用户名密码，迁移数据
-3. **已有账户** → 直接登录进入聊天界面
-
-> 💡 所有用户数据隔离存储：`~/.trulymem/{用户名}/`
-
-### 聊天配置
-
-登录后按 **F2** 展开右侧配置面板：
-
-1. **API Key** — 必须（支持 DeepSeek、OpenAI 等）
-2. **模型** — 可选，默认已配置
-3. **Base URL** — 可选
-
-配置自动保存，下次启动自动加载。
-
---
-
-## Web 可视化界面
-
-TrulyMEM 的 Web 服务现在**内嵌在主进程中**（无需独立启动子进程）。
-
-### TUI 内启动（管理员专有）
-
-管理员按 F2 打开右侧面板，勾选「启用 Web 服务」即可。
-
-### 手动启动
-
-```bash
-python -m core.web_api --port 4096
-# 访问 http://localhost:4096
-```
-
-### 首次访问流程
-
-1. 浏览器打开 `http://localhost:4096`
-2. **无用户** → 自动跳转至设置页，创建管理员账号
-3. **有用户** → 跳转至登录页
-4. 登录后进入星图可视化页面
-
-### Web 功能
-
-| 页面 | 访问权限 | 功能 |
-|------|----------|------|
-| 🌟 星图 | 所有已登录用户 | 浏览知识图谱三元组 |
-| ⚙ 设置 | 所有已登录用户 | 修改密码 |
-| 🧑‍💼 用户管理 | **仅管理员** | 添加/删除用户 |
-
---
-
-## 多用户系统
-
-### 目录结构
-
-```
-~/.trulymem/
-├── trulymem.db          # 全局用户数据库（web_users 表）
-├── .migrated            # 旧版迁移标记
-├── admin/
-│   ├── config.json      # 管理员配置
-│   └── admin_graph.db   # 管理员知识图谱
-└── user2/
-    ├── config.json      # user2 配置
-    └── user2_graph.db   # user2 知识图谱
-```
-
-### 角色体系
-
-| 功能 | 普通用户 | 管理员 |
-|------|---------|--------|
-| 修改自己密码 | ✅ | ✅ |
-| 配置 API Key / 模型 | ✅ | ✅ |
-| Web 服务开关（TUI） | ❌ | ✅ |
-| Web 登录凭据 | ❌ | ✅ |
-| 查看用户列表 | ❌ | ✅ |
-| 添加/删除用户 | ❌ | ✅ |
-
-> ⚠️ 首个注册用户自动成为管理员。Web 设置页可添加新用户。
-
---
-
-## 快捷键
-
-| 按键 | 功能 |
-|------|------|
-| F1 | 帮助 |
-| F2 | 切换侧边栏（配置面板） |
-| F3 | 工具详情 |
-| F5 | 清屏 |
-| F6 | 退出 |
-
---
-
-## 打包构建
-
-```bash
-# 安装依赖
-pip install -r requirements.txt
-
-# 构建（Linux / macOS）
-pyinstaller --clean build/trulymem.spec
-```
-
-构建产出：`dist/TrulyMEM`（单文件，TUI + Web 服务均内嵌于同一二进制）
-
-> 📦 从 v2 开始，Web 服务作为线程嵌入主程序，不再需要独立打包 `trulymem-web`。
-> 
-> 💡 修改 `ui/static/` 或 `core/` 等源码后必须重新编译才能生效（静态文件在构建时打入二进制）。
-
---
-
-## 架构说明
-
-### 通信协议
-
-UI 与后端通过 **Packet 协议** 通信：
-
-```
-TUI (Textual)  ←→  BackendClient  ←→  queue.Queue  ←→  BackendServer (独立线程)
-```
-
-### 配置管理
-
- **用户级存储**: `~/.trulymem/{username}/config.json`
- **Web 配置**: `~/.trulymem/trulymem.db`（web_users 表）
- **自动加载**: 启动时根据登录用户加载对应配置文件
- **动态更新**: 运行时修改配置立即生效，自动持久化
-
-### Web 服务架构
-
-```
-┌──────────────────────┐
-│    TrulyMEM 主进程    │
-│  ┌──────┐ ┌────────┐ │
-│  │ TUI  │ │ Flask  │ │  ← 同一进程，不同线程
-│  │      │ │ Thread │ │
-│  └──────┘ └────────┘ │
-└──────────────────────┘
-```
-
---
-
-## 常见问题
-
-### Python 未找到
-
-安装 Python 3.8+：https://www.python.org/downloads/
-
-### 依赖安装失败
-
-```bash
-python -m venv venv
-source venv/bin/activate  # Linux/macOS
-venv\Scripts\activate     # Windows
-pip install -r requirements.txt
-```
-
-### API Key 无效
-
-检查 API Key 格式，确保无多余空格。可在 TUI 右侧面板重新配置。
-
-### 管理员账号丢失
-
-数据库中第一个注册账号总是 admin。如果所有用户都丢失了 admin 权限，删除用户目录下的 `trulymem.db` 后重新注册即可。
-
-### 旧版数据迁移
-
-检测到旧版 `~/.trulymem/config.json` 和 `graph_memory.db` 时，TUI 启动时自动进入迁移引导。迁移后旧文件保留，不会删除。
-
---
-
-## 开发命令
-
-```bash
-pip install -r requirements.txt
-pytest tests/
-bash build/build_linux.sh
-```
--- a/docs/zh/working_memory.md
+++ b/docs/zh/working_memory.md
@ -1,182 +0,0 @@
-# 工作记忆链机制说明
-
-## 概述
-
-TrulyMEM 通过工作记忆链机制维持对话连贯性。由于系统没有传统的消息历史数组，图数据库是唯一的记忆载体，工作记忆链是维持对话上下文的关键机制。
-
-## 核心问题
-
-传统 AI 对话系统在处理连续性任务时存在以下问题:
-
-1. **没有工作记忆链**: AI 无法记住当前正在进行的任务状态
-2. **任务上下文丢失**: 当话题被打断后，AI 无法恢复之前的任务
-3. **缺乏任务状态管理**: 没有明确标注任务的完成状态
-
-### 问题示例
-
-```
-用户: 咱来玩成语接龙吧，我先开始，为所欲为
-AI: 好的喵！我接：为虎作伥喵！
-
-用户: 长门有希  (话题被打断)
-AI: (讨论长门有希的内容)
-
-用户: 关于刚才的成语接龙，我并不知道应该怎么接你的成语，请帮我接一下
-AI: [猜测] 看起来我们之前应该没有进行过成语接龙游戏...
-```
-
-**问题**: AI 完全忘记了之前的成语接龙游戏。
-
-## 解决方案
-
-### 专用工具
-
-系统提供 4 个专用任务工具:
-
-| 工具 | 功能 | 使用场景 |
-|------|------|----------|
-| `task_create` | 创建任务节点 | 开始新任务 |
-| `task_set_state` | 设置任务状态 | 更新进行中/已完成/已暂停/已取消 |
-| `task_delete` | 删除任务 | 清理完成任务 |
-| `task_link_info` | 关联信息节点 | 连接任务与具体信息 |
-
-### 任务状态
-
- **进行中**: 任务正在执行
- **已完成**: 任务成功完成
- **已暂停**: 任务被中断，可恢复
- **已取消**: 任务被取消
-
-## 使用流程
-
-### 每轮对话必须执行
-
-1. **查询人设图** (最高优先级)
-   ```
-   调用 memory_recall
-   参数: {"query_intent": "AI,人设,角色,性格,语气,说话风格", "depth": 2}
-   ```
-
-2. **查询工作记忆链**
-   ```
-   调用 memory_recall
-   参数: {"query_intent": "TaskNode,工作记忆,任务链", "depth": 2}
-   ```
-
-3. **根据上下文生成回复**
-
-4. **更新工作记忆链** (如有必要)
-
-## 完整示例: 成语接龙游戏
-
-### 第一轮: 用户发起游戏
-
-```
-用户: 咱来玩成语接龙吧，我先开始，为所欲为
-
-AI操作:
-1. 查询人设图 → 获取当前人设（如：猫娘）
-2. 查询工作记忆链 → 无进行中任务
-3. 记录游戏状态:
-   调用 memory_commit
-   参数: {
-     "triplets": [
-       {"subject": "成语接龙_当前成语", "relation": "内容", "object": "为所欲为"},
-       {"subject": "成语接龙_当前成语", "relation": "游戏", "object": "成语接龙"}
-     ]
-   }
-4. 创建任务节点:
-   调用 task_create
-   参数: {
-     "task_id": "Task_成语接龙",
-     "description": "成语接龙游戏，当前成语：为所欲为",
-     "info_nodes": ["成语接龙_当前成语"]
-   }
-5. 回复: "好的喵！我接：为虎作伥喵！"
-```
-
-### 第二轮: 话题被打断
-
-```
-用户: 长门有希
-
-AI操作:
-1. 查询人设图 → 获取当前人设（猫娘）
-2. 查询工作记忆链 → 发现 Task_成语接龙 状态为"进行中"
-3. 暂停任务:
-   调用 task_set_state
-   参数: {"task_id": "Task_成语接龙", "state": "已暂停"}
-4. 创建新任务:
-   调用 task_create
-   参数: {"task_id": "Task_长门有希", "description": "讨论长门有希"}
-5. 回复关于长门有希的内容
-```
-
-### 第三轮: 用户要求继续游戏
-
-```
-用户: 关于刚才的成语接龙，我并不知道应该怎么接你的成语，请帮我接一下
-
-AI操作:
-1. 查询人设图 → 获取当前人设（猫娘）
-2. 查询工作记忆链 → 发现 Task_成语接龙 状态为"已暂停"
-3. 恢复任务:
-   调用 task_set_state
-   参数: {"task_id": "Task_成语接龙", "state": "进行中"}
-4. 查询信息节点 → 获取当前成语"为虎作伥"
-5. 回复: "好的喵！上一个成语是'为虎作伥'，我帮你接：伥鬼害人喵！"
-```
-
-## API 参考
-
-### task_create
-
-创建任务节点，用于跟踪连续性任务。
-
-```json
-{
-  "task_id": "Task_成语接龙",
-  "description": "任务概述",
-  "info_nodes": ["关联的信息节点名称"]
-}
-```
-
-### task_set_state
-
-设置任务状态。
-
-```json
-{
-  "task_id": "Task_成语接龙",
-  "state": "进行中"  // 进行中/已完成/已暂停/已取消
-}
-```
-
-### task_delete
-
-删除任务节点。
-
-```json
-{
-  "task_id": "Task_成语接龙",
-  "delete_info_nodes": true  // 是否删除关联的信息节点
-}
-```
-
-### task_link_info
-
-关联信息节点到任务。
-
-```json
-{
-  "task_id": "Task_成语接龙",
-  "info_node_names": ["成语接龙_当前成语", "成语接龙_上一个成语"]
-}
-```
-
-## 注意事项
-
-1. **人设图优先级最高**: 每轮必须首先查询人设图
-2. **工作记忆链是唯一上下文载体**: 没有传统消息历史
-3. **任务状态必须及时更新**: 确保状态转换正确
-4. **使用专用工具**: 优先使用 task_* 工具而非 memory_commit 处理任务相关操作
--- a/features/chat/BuildProfile.ets
+++ b/features/chat/BuildProfile.ets
@ -0,0 +1,17 @@
+/**
+ * Use these variables when you tailor your ArkTS code. They must be of the const type.
+ */
+export const HAR_VERSION = '1.0.0';
+export const BUILD_MODE_NAME = 'debug';
+export const DEBUG = true;
+export const TARGET_NAME = 'default';
+
+/**
+ * BuildProfile Class is used only for compatibility purposes.
+ */
+export default class BuildProfile { 
+	static readonly HAR_VERSION = HAR_VERSION;
+	static readonly BUILD_MODE_NAME = BUILD_MODE_NAME;
+	static readonly DEBUG = DEBUG;
+	static readonly TARGET_NAME = TARGET_NAME;
+}
--- a/features/chat/Index.ets
+++ b/features/chat/Index.ets
@ -0,0 +1 @@
+export { ChatPage } from './src/main/ets/pages/ChatPage';
--- a/features/chat/build-profile.json5
+++ b/features/chat/build-profile.json5
@ -0,0 +1,10 @@
+{
+  "apiType": "stageMode",
+  "buildOption": {
+  },
+  "targets": [
+    {
+      "name": "default"
+    }
+  ]
+}
--- a/features/chat/build/default/intermediates/merge_profile/default/module.json
+++ b/features/chat/build/default/intermediates/merge_profile/default/module.json
@ -1,30 +0,0 @@
-{
-	"app": {
-		"bundleName": "com.trulymem.app",
-		"debug": true,
-		"versionCode": 1000001,
-		"versionName": "1.0.0",
-		"minAPIVersion": 60100023,
-		"targetAPIVersion": 60100023,
-		"apiReleaseType": "Release",
-		"targetMinorAPIVersion": 0,
-		"targetPatchAPIVersion": 0,
-		"compileSdkVersion": "6.1.0.105",
-		"compileSdkType": "HarmonyOS",
-		"appEnvironments": [],
-		"bundleType": "app",
-		"buildMode": "debug"
-	},
-	"module": {
-		"name": "chat",
-		"type": "har",
-		"description": "TrulyMEM chat feature module",
-		"deviceTypes": [
-			"phone",
-			"tablet",
-			"2in1"
-		],
-		"packageName": "@ohos/chat",
-		"installationFree": false
-	}
-}
--- a/features/chat/chat
+++ b/features/chat/chat
@ -0,0 +1 @@
+/home/program/TrulyMEM-TrueHumanMEM/features/chat
--- a/features/chat/hvigorfile.ts
+++ b/features/chat/hvigorfile.ts
@ -0,0 +1,6 @@
+import { harTasks } from '@ohos/hvigor-ohos-plugin';
+
+export default {
+    system: harTasks,
+    plugins: []
+};
--- a/features/chat/oh-package-lock.json5
+++ b/features/chat/oh-package-lock.json5
@ -0,0 +1,19 @@
+{
+  "meta": {
+    "stableOrder": true,
+    "enableUnifiedLockfile": false
+  },
+  "lockfileVersion": 3,
+  "ATTENTION": "THIS IS AN AUTOGENERATED FILE. DO NOT EDIT THIS FILE DIRECTLY.",
+  "specifiers": {
+    "@ohos/common@../../common": "@ohos/common@../../common"
+  },
+  "packages": {
+    "@ohos/common@../../common": {
+      "name": "@ohos/common",
+      "version": "1.0.0",
+      "resolved": "../../common",
+      "registryType": "local"
+    }
+  }
+}
--- a/features/chat/oh-package.json5
+++ b/features/chat/oh-package.json5
@ -0,0 +1,11 @@
+{
+  "name": "@ohos/chat",
+  "version": "1.0.0",
+  "description": "TrulyMEM chat feature module",
+  "main": "Index.ets",
+  "author": "",
+  "license": "",
+  "dependencies": {
+    "@ohos/common": "file:../../common"
+  }
+}
--- a/features/chat/oh_modules/@ohos/common
+++ b/features/chat/oh_modules/@ohos/common
@ -0,0 +1 @@
+../../../../common
--- a/features/chat/src/main/ets/components/ChatComponents.ets
+++ b/features/chat/src/main/ets/components/ChatComponents.ets
@ -0,0 +1,161 @@
+import { ChatMessage } from '@ohos/common';
+
+/**
+ * ChatMessageBubble — 单条聊天消息气泡
+ * 封装消息的角色标识、内容样式、玻璃拟态背景
+ */
+@Component
+export struct ChatMessageBubble {
+  @ObjectLink msg: ChatMessage;
+
+  build() {
+    Column() {
+      // 角色标识
+      Text(this.msg.role === 'user' ? '🧑 你' : '🤖 AI')
+        .fontSize(11)
+        .fontColor(this.msg.role === 'user' ? '#7C4DFF' : '#999')
+        .width('100%')
+
+      // 消息内容
+      Text(this.msg.content)
+        .fontSize(15)
+        .width('100%')
+        .margin({ top: 4 })
+        .fontColor('#FFFFFF')
+    }
+    .padding(12)
+    .backgroundColor(this.msg.role === 'user' ? 'rgba(124,77,255,0.15)' : 'rgba(245,245,245,0.1)')
+    .borderRadius(12)
+    .border({
+      width: 1,
+      color: this.msg.role === 'user' ? 'rgba(124,77,255,0.3)' : 'rgba(255,255,255,0.1)'
+    })
+    .backgroundBlurStyle(BlurStyle.Thin)
+    .margin({ left: 8, right: 8, bottom: 8 })
+    .width('100%')
+    .alignItems(HorizontalAlign.Start)
+  }
+}
+
+/**
+ * ThinkingIndicator — AI 思考中指示器
+ * 玻璃拟态加载动画 + 文字提示
+ */
+@Component
+export struct ThinkingIndicator {
+  build() {
+    Row() {
+      LoadingProgress()
+        .width(20)
+        .height(20)
+        .margin({ right: 8 })
+        .color('#7C4DFF')
+      Text('AI 思考中...')
+        .fontSize(13)
+        .fontColor('#7C4DFF')
+    }
+    .padding(12)
+    .backgroundColor('rgba(124,77,255,0.1)')
+    .borderRadius(12)
+    .border({ width: 1, color: 'rgba(124,77,255,0.2)' })
+    .backgroundBlurStyle(BlurStyle.Thin)
+    .margin({ left: 8, right: 8, bottom: 8 })
+  }
+}
+
+/**
+ * ToolCallLogPanel — 工具调用日志面板
+ * 橙色风格，显示 Agent 调用的工具链
+ */
+@Component
+export struct ToolCallLogPanel {
+  @Prop logText: string;
+
+  build() {
+    Text(this.logText)
+      .fontSize(10)
+      .fontColor('#FF9800')
+      .backgroundColor('rgba(255,152,0,0.1)')
+      .padding(8)
+      .borderRadius(8)
+      .border({ width: 1, color: 'rgba(255,152,0,0.2)' })
+      .backgroundBlurStyle(BlurStyle.Thin)
+      .margin({ left: 8, right: 8, bottom: 4 })
+      .lineHeight(16)
+  }
+}
+
+/**
+ * ChatInputBar — 底部输入栏
+ * TextArea + 发送按钮，主题色边框
+ */
+@Component
+export struct ChatInputBar {
+  @Link inputText: string;
+  @Prop isThinking: boolean;
+  onSend?: () => void;
+
+  build() {
+    Row() {
+      TextArea({ text: this.inputText, placeholder: '输入消息...' })
+        .layoutWeight(1)
+        .onChange((v: string) => { this.inputText = v; })
+        .height(40)
+        .backgroundColor('rgba(255,255,255,0.1)')
+        .borderRadius(8)
+        .border({ width: 1, color: 'rgba(124,77,255,0.3)' })
+
+      Button('发送')
+        .enabled(!this.isThinking)
+        .onClick(() => { this.onSend?.(); })
+        .backgroundColor('#7C4DFF')
+        .borderRadius(8)
+    }
+    .width('100%')
+    .padding(8)
+    .backgroundColor('rgba(255,255,255,0.05)')
+    .backgroundBlurStyle(BlurStyle.Regular)
+    .border({
+      width: 1,
+      color: 'rgba(124,77,255,0.2)',
+      style: BorderStyle.Solid
+    })
+  }
+}
+
+/**
+ * ChatMessageList — 聊天消息列表容器
+ * 整合消息气泡、思考指示器、工具日志
+ */
+@Component
+export struct ChatMessageList {
+  @Prop messages: ChatMessage[];
+  @Prop isThinking: boolean;
+  @Prop toolCallLog: string;
+  private scrollController: Scroller = new Scroller();
+
+  build() {
+    List() {
+      ForEach(this.messages, (msg: ChatMessage) => {
+        ListItem() {
+          ChatMessageBubble({ msg: msg })
+        }
+      })
+
+      if (this.isThinking) {
+        ListItem() {
+          ThinkingIndicator()
+        }
+      }
+
+      if (this.toolCallLog && !this.isThinking) {
+        ListItem() {
+          ToolCallLogPanel({ logText: this.toolCallLog })
+        }
+      }
+    }
+    .width('100%')
+    .layoutWeight(1)
+    .backgroundColor('rgba(0,0,0,0.1)')
+  }
+}
--- a/features/chat/src/main/ets/pages/ChatPage.ets
+++ b/features/chat/src/main/ets/pages/ChatPage.ets
@ -0,0 +1,88 @@
+import { GraphDatabase, GraphMemoryService, AIAgentService, ChatMessage, AgentResponse, Logger } from '@ohos/common';
+import { ChatMessageList, ChatInputBar } from '../components/ChatComponents';
+
+@Component
+export struct ChatPage {
+    @State messages: ChatMessage[] = [];
+    @State inputText: string = '';
+    @Prop db: GraphDatabase;
+    @State toolCallLog: string = '';
+    @State isThinking: boolean = false;
+    private agentService?: AIAgentService;
+
+    async aboutToAppear() {
+        // 初始化图记忆服务和 Agent
+        const memoryService = new GraphMemoryService(this.db);
+        this.agentService = new AIAgentService(memoryService, getContext(this));
+
+        // 加载历史消息（兼容旧数据：无 session_id 时加载全部）
+        const rawHistory = await this.db.getChatHistory(50, this.agentService.getSessionId());
+        if (rawHistory.length === 0) {
+            // 新 session，尝试加载旧消息
+            const legacyHistory = await this.db.getChatHistory(50);
+            this.messages = legacyHistory.map(m => {
+                const msg: ChatMessage = { role: m.role, content: m.content };
+                return msg;
+            });
+        } else {
+            this.messages = rawHistory.map(m => {
+                const msg: ChatMessage = { role: m.role, content: m.content };
+                return msg;
+            });
+        }
+    }
+
+    async sendMessage() {
+        if (!this.inputText.trim() || !this.agentService) return;
+
+        const userMessage: string = this.inputText;
+        this.inputText = '';
+
+        // 添加用户消息
+        await this.db.saveChatMessage('user', userMessage, '', this.agentService.getSessionId());
+        this.messages = [...this.messages, { role: 'user', content: userMessage }];
+
+        // 显示 loading
+        this.isThinking = true;
+        this.toolCallLog = '';
+
+        try {
+            // 通过 Agent 发送消息
+            const agentResponse: AgentResponse = await this.agentService.sendMessage(userMessage);
+
+            // 记录工具调用日志
+            if (agentResponse.toolCalls.length > 0) {
+                const logs: string[] = agentResponse.toolCalls.map(tc => `🛠 ${tc.name}: ${tc.message}`);
+                this.toolCallLog = logs.join('\n');
+            }
+
+            // 保存并显示 AI 回复
+            await this.db.saveChatMessage('assistant', agentResponse.content, this.toolCallLog, this.agentService.getSessionId());
+            this.messages = [...this.messages, { role: 'assistant', content: agentResponse.content }];
+        } catch (err) {
+            Logger.error('Agent request failed: ' + JSON.stringify(err));
+            this.messages = [...this.messages, { role: 'assistant', content: `⚠️ 请求失败: ${err.message || JSON.stringify(err)}` }];
+        } finally {
+            this.isThinking = false;
+        }
+    }
+
+    build() {
+        Column() {
+            ChatMessageList({
+                messages: this.messages,
+                isThinking: this.isThinking,
+                toolCallLog: this.toolCallLog
+            })
+
+            ChatInputBar({
+                inputText: this.inputText,
+                isThinking: this.isThinking,
+                onSend: (): void => { this.sendMessage(); }
+            })
+        }
+        .width('100%')
+        .height('100%')
+        .backgroundColor('rgba(26,27,46,0.95)')
+    }
+}
--- a/features/chat/src/main/module.json5
+++ b/features/chat/src/main/module.json5
@ -0,0 +1,12 @@
+{
+  "module": {
+    "name": "chat",
+    "type": "har",
+    "description": "TrulyMEM chat feature module",
+    "deviceTypes": [
+      "phone",
+      "tablet",
+      "2in1"
+    ]
+  }
+}
--- a/features/graph/BuildProfile.ets
+++ b/features/graph/BuildProfile.ets
@ -0,0 +1,17 @@
+/**
+ * Use these variables when you tailor your ArkTS code. They must be of the const type.
+ */
+export const HAR_VERSION = '1.0.0';
+export const BUILD_MODE_NAME = 'debug';
+export const DEBUG = true;
+export const TARGET_NAME = 'default';
+
+/**
+ * BuildProfile Class is used only for compatibility purposes.
+ */
+export default class BuildProfile { 
+	static readonly HAR_VERSION = HAR_VERSION;
+	static readonly BUILD_MODE_NAME = BUILD_MODE_NAME;
+	static readonly DEBUG = DEBUG;
+	static readonly TARGET_NAME = TARGET_NAME;
+}
--- a/features/graph/Index.ets
+++ b/features/graph/Index.ets
@ -0,0 +1 @@
+export { GraphPage } from './src/main/ets/pages/GraphPage';
--- a/features/graph/build-profile.json5
+++ b/features/graph/build-profile.json5
@ -0,0 +1,10 @@
+{
+  "apiType": "stageMode",
+  "buildOption": {
+  },
+  "targets": [
+    {
+      "name": "default"
+    }
+  ]
+}
--- a/features/graph/build/default/intermediates/merge_profile/default/module.json
+++ b/features/graph/build/default/intermediates/merge_profile/default/module.json
@ -1,30 +0,0 @@
-{
-	"app": {
-		"bundleName": "com.trulymem.app",
-		"debug": true,
-		"versionCode": 1000001,
-		"versionName": "1.0.0",
-		"minAPIVersion": 60100023,
-		"targetAPIVersion": 60100023,
-		"apiReleaseType": "Release",
-		"targetMinorAPIVersion": 0,
-		"targetPatchAPIVersion": 0,
-		"compileSdkVersion": "6.1.0.105",
-		"compileSdkType": "HarmonyOS",
-		"appEnvironments": [],
-		"bundleType": "app",
-		"buildMode": "debug"
-	},
-	"module": {
-		"name": "graph",
-		"type": "har",
-		"description": "TrulyMEM graph feature module",
-		"deviceTypes": [
-			"phone",
-			"tablet",
-			"2in1"
-		],
-		"packageName": "@ohos/graph",
-		"installationFree": false
-	}
-}
--- a/features/graph/graph
+++ b/features/graph/graph
@ -0,0 +1 @@
+/home/program/TrulyMEM-TrueHumanMEM/features/graph
--- a/features/graph/hvigorfile.ts
+++ b/features/graph/hvigorfile.ts
@ -0,0 +1,6 @@
+import { harTasks } from '@ohos/hvigor-ohos-plugin';
+
+export default {
+    system: harTasks,
+    plugins: []
+};
--- a/features/graph/oh-package-lock.json5
+++ b/features/graph/oh-package-lock.json5
@ -0,0 +1,19 @@
+{
+  "meta": {
+    "stableOrder": true,
+    "enableUnifiedLockfile": false
+  },
+  "lockfileVersion": 3,
+  "ATTENTION": "THIS IS AN AUTOGENERATED FILE. DO NOT EDIT THIS FILE DIRECTLY.",
+  "specifiers": {
+    "@ohos/common@../../common": "@ohos/common@../../common"
+  },
+  "packages": {
+    "@ohos/common@../../common": {
+      "name": "@ohos/common",
+      "version": "1.0.0",
+      "resolved": "../../common",
+      "registryType": "local"
+    }
+  }
+}
--- a/features/graph/oh-package.json5
+++ b/features/graph/oh-package.json5
@ -0,0 +1,11 @@
+{
+  "name": "@ohos/graph",
+  "version": "1.0.0",
+  "description": "TrulyMEM graph feature module",
+  "main": "Index.ets",
+  "author": "",
+  "license": "",
+  "dependencies": {
+    "@ohos/common": "file:../../common"
+  }
+}
--- a/features/graph/oh_modules/@ohos/common
+++ b/features/graph/oh_modules/@ohos/common
@ -0,0 +1 @@
+../../../../common
--- a/features/graph/src/main/ets/components/GraphComponents.ets
+++ b/features/graph/src/main/ets/components/GraphComponents.ets
@ -0,0 +1,251 @@
+import web_webview from '@ohos.web.webview';
+import { GraphDatabase, RecallEntity, GraphMemoryService, ConnectionItem, NodeDetailInfo, Logger } from '@ohos/common';
+
+/**
+ * GraphNodeSearchBar — 图节点搜索栏
+ * 悬浮在 WebView 上方的搜索输入框
+ */
+@Component
+export struct GraphNodeSearchBar {
+  @Link searchText: string;
+  onSearchInput?: (value: string) => void;
+
+  build() {
+    Column() {
+      TextInput({ placeholder: '搜索节点...', text: this.searchText })
+        .width('80%')
+        .height(40)
+        .backgroundColor('rgba(10, 10, 26, 0.8)')
+        .fontColor('#ffffff')
+        .placeholderColor('#666688')
+        .borderRadius(8)
+        .border({ width: 1, color: 'rgba(100, 100, 255, 0.3)' })
+        .margin({ top: 20 })
+        .onChange((value: string) => {
+          this.onSearchInput?.(value);
+        })
+    }
+    .width('100%')
+    .position({ x: 0, y: 0 })
+    .zIndex(10)
+  }
+}
+
+/**
+ * NodeDetailPanel — 节点详情浮层
+ * 显示选中节点的名称、类型、提及次数、连接关系
+ */
+@Component
+export struct NodeDetailPanel {
+  @Prop detail: NodeDetailInfo;
+  onClose?: () => void;
+
+  build() {
+    Column() {
+      Column() {
+        Text(this.detail.name)
+          .fontSize(18)
+          .fontColor('#44ff88')
+          .fontWeight(FontWeight.Bold)
+          .margin({ bottom: 10 })
+
+        Text('类型: ' + this.detail.type)
+          .fontSize(14)
+          .fontColor('#aaaacc')
+
+        Text('提及次数: ' + this.detail.mention_count)
+          .fontSize(14)
+          .fontColor('#aaaacc')
+
+        Text('连接数: ' + this.detail.connection_count)
+          .fontSize(14)
+          .fontColor('#aaaacc')
+
+        if (this.detail.connections && this.detail.connections.length > 0) {
+          Text('连接关系:')
+            .fontSize(14)
+            .fontColor('#8888aa')
+            .margin({ top: 10, bottom: 5 })
+          List() {
+            ForEach(this.detail.connections, (conn: ConnectionItem) => {
+              ListItem() {
+                Text(conn.type + ': ' + conn.target_name)
+                  .fontSize(12)
+                  .fontColor('#aaaacc')
+              }
+            })
+          }
+          .height(100)
+        }
+
+        Button('关闭')
+          .width(80)
+          .height(30)
+          .margin({ top: 15 })
+          .backgroundColor('rgba(100, 100, 255, 0.3)')
+          .fontColor('#ffffff')
+          .onClick(() => {
+            this.onClose?.();
+          })
+      }
+      .padding(20)
+      .backgroundColor('rgba(10, 10, 26, 0.95)')
+      .borderRadius(12)
+      .border({ width: 1, color: 'rgba(100, 100, 255, 0.3)' })
+      .width(300)
+    }
+    .width('100%')
+    .height('100%')
+    .backgroundColor('rgba(0, 0, 0, 0.5)')
+    .justifyContent(FlexAlign.Center)
+    .alignItems(HorizontalAlign.Center)
+    .zIndex(20)
+  }
+}
+
+/**
+ * GraphWebView — 图可视化 WebView 封装
+ * 包含 WebView 配置、JS Bridge 注册、数据加载回调
+ */
+@Component
+export struct GraphWebView {
+  private controller: web_webview.WebviewController = new web_webview.WebviewController();
+  private bridge?: NativeBridge;
+
+  onPageEnd?: () => void;
+
+  getController(): web_webview.WebviewController {
+    return this.controller;
+  }
+
+  setBridge(bridge: NativeBridge): void {
+    this.bridge = bridge;
+  }
+
+  build() {
+    Web({ src: $rawfile('graph.html'), controller: this.controller })
+      .javaScriptAccess(true)
+      .width('100%')
+      .height('100%')
+      .zoomAccess(true)
+      .onPageEnd(() => {
+        this.onPageEnd?.();
+      })
+      .javaScriptProxy({
+        object: this.bridge,
+        name: 'nativeBridge',
+        methodList: ['onNodeClick', 'onSearch'],
+        asyncMethodList: ['requestGraphData'],
+        controller: this.controller
+      })
+  }
+}
+
+/**
+ * NativeBridge — WebView 原生桥接类（移动自 GraphPage）
+ * 负责 ArkTS ↔ WebView JavaScript 双向通信
+ */
+export class NativeBridge {
+  private controller: web_webview.WebviewController;
+  private onRequestGraphData: () => void;
+  private onNodeClickCallback: (nodeId: number, nodeName: string) => void;
+  private onSearchCallback: (query: string) => void;
+
+  constructor(
+    controller: web_webview.WebviewController,
+    onRequestGraphData: () => void,
+    onNodeClickCallback: (nodeId: number, nodeName: string) => void,
+    onSearchCallback: (query: string) => void
+  ) {
+    this.controller = controller;
+    this.onRequestGraphData = onRequestGraphData;
+    this.onNodeClickCallback = onNodeClickCallback;
+    this.onSearchCallback = onSearchCallback;
+  }
+
+  onNodeClick(nodeId: number, nodeName: string): void {
+    Logger.info('Node clicked: id=' + nodeId + ', name=' + nodeName);
+    if (this.onNodeClickCallback) {
+      this.onNodeClickCallback(nodeId, nodeName);
+    }
+  }
+
+  onSearch(query: string): void {
+    Logger.info('Search from WebView: ' + query);
+    if (this.onSearchCallback) {
+      this.onSearchCallback(query);
+    }
+  }
+
+  requestGraphData(): void {
+    Logger.info('requestGraphData called from WebView');
+    if (this.onRequestGraphData) {
+      this.onRequestGraphData();
+    }
+  }
+}
+
+/**
+ * GraphDataService — 图数据查询服务
+ * 封装从 GraphDatabase 读取节点和边的逻辑
+ */
+export class GraphDataService {
+  private db: GraphDatabase;
+
+  constructor(db: GraphDatabase) {
+    this.db = db;
+  }
+
+  async getAllNodes(): Promise<GraphNodeItem[]> {
+    const result = await this.db.search('');
+    return result.map((r, idx): GraphNodeItem => {
+      return {
+        id: idx + 1,
+        label: r.name,
+        type: r.type,
+        mentions: r.mentions
+      };
+    });
+  }
+
+  async getAllEdges(): Promise<GraphEdgeItem[]> {
+    const recallResult = await this.db.recall('', [], 3);
+    const nameToId: Record<string, number> = {};
+    recallResult.entities.forEach((e: RecallEntity, idx: number): void => {
+      nameToId[e.name as string] = idx + 1;
+    });
+    const edgeItems: GraphEdgeItem[] = [];
+    for (let i = 0; i < recallResult.relations.length; i++) {
+      const r = recallResult.relations[i];
+      const sourceId = nameToId[r.source];
+      const targetId = nameToId[r.target];
+      if (sourceId !== undefined && targetId !== undefined) {
+        edgeItems.push({
+          id: i + 1,
+          source: sourceId,
+          target: targetId,
+          label: r.type,
+          relation: r.type
+        });
+      }
+    }
+    return edgeItems;
+  }
+}
+
+// ========= 内部类型定义 =========
+
+interface GraphNodeItem {
+  id: number;
+  label: string;
+  type: string;
+  mentions: number;
+}
+
+interface GraphEdgeItem {
+  id: number;
+  source: number;
+  target: number;
+  label: string;
+  relation: string;
+}
--- a/features/graph/src/main/ets/pages/GraphPage.ets
+++ b/features/graph/src/main/ets/pages/GraphPage.ets
@ -0,0 +1,200 @@
+/**
+ * GraphPage — 记忆星图页面（重构后）
+ * 使用 WebView 显示 Three.js 3D 图可视化
+ * 子组件：GraphWebView、GraphNodeSearchBar、NodeDetailPanel、GraphDataService、NativeBridge
+ */
+import web_webview from '@ohos.web.webview';
+import {
+  GraphDatabase,
+  NodeDetailInfo,
+  Logger,
+  RecallEntity,
+  GraphMemoryService
+} from '@ohos/common';
+import {
+  GraphWebView,
+  GraphNodeSearchBar,
+  NodeDetailPanel,
+  GraphDataService,
+  NativeBridge
+} from '../components/GraphComponents';
+
+// ========= GraphPage 组件 =========
+
+interface GraphNodeItem {
+    id: number;
+    label: string;
+    type: string;
+    mentions: number;
+}
+
+interface GraphEdgeItem {
+    id: number;
+    source: number;
+    target: number;
+    label: string;
+    relation: string;
+}
+
+@Component
+export struct GraphPage {
+    private controller: web_webview.WebviewController = new web_webview.WebviewController();
+    @Prop db: GraphDatabase;
+    @State nodeCount: number = 0;
+    @State edgeCount: number = 0;
+    @State selectedNodeDetail: NodeDetailInfo | null = null;
+    @State showNodeDetail: boolean = false;
+    @State searchText: string = '';
+    private graphService: GraphMemoryService = new GraphMemoryService(this.db);
+
+    // 初始化桥接对象
+    private bridge: NativeBridge = new NativeBridge(
+        this.controller,
+        (): void => { this.pushGraphDataToWebView(); },
+        (nodeId: number, nodeName: string): void => { this.handleNodeClick(nodeId, nodeName); },
+        (query: string): void => { this.handleSearchFromWeb(query); }
+    );
+
+    /**
+     * 外部触发刷新图数据（聊天写入新记忆后调用）
+     */
+    public async refreshGraphData(): Promise<void> {
+        await this.pushGraphDataToWebView();
+    }
+
+    /**
+     * 处理节点点击 - 查询详细信息并显示浮层
+     */
+    private async handleNodeClick(nodeId: number, nodeName: string): Promise<void> {
+        try {
+            const detail: NodeDetailInfo | null = await this.graphService.getNodeDetail(nodeName);
+            if (detail) {
+                this.selectedNodeDetail = detail;
+                this.showNodeDetail = true;
+            }
+        } catch (err) {
+            Logger.error('handleNodeClick error: ' + JSON.stringify(err));
+        }
+    }
+
+    /**
+     * 处理来自 WebView 的搜索请求
+     */
+    private handleSearchFromWeb(query: string): void {
+        this.searchText = query;
+    }
+
+    /**
+     * 处理搜索输入 - 通知 WebView 过滤
+     */
+    private onSearchInput(value: string): void {
+        this.searchText = value;
+        const jsCode = `window.dispatchEvent(new MessageEvent('message', { data: { type: 'search_nodes', query: '${value}' } }));`;
+        this.controller.runJavaScript(jsCode);
+    }
+
+    /**
+     * 关闭节点详情浮层
+     */
+    private closeNodeDetail(): void {
+        this.showNodeDetail = false;
+        this.selectedNodeDetail = null;
+    }
+
+    /**
+     * 从数据库读取全量图数据，推送给 WebView
+     */
+    private async getAllNodesData(): Promise<GraphNodeItem[]> {
+        const result = await this.db.search('');
+        return result.map((r, idx): GraphNodeItem => {
+            return {
+                id: idx + 1,
+                label: r.name,
+                type: r.type,
+                mentions: r.mentions
+            };
+        });
+    }
+
+    private async getAllEdgesData(): Promise<GraphEdgeItem[]> {
+        const recallResult = await this.db.recall('', [], 3);
+        const nameToId: Record<string, number> = {};
+        recallResult.entities.forEach((e: RecallEntity, idx: number): void => {
+            nameToId[e.name as string] = idx + 1;
+        });
+        const edgeItems: GraphEdgeItem[] = [];
+        for (let i = 0; i < recallResult.relations.length; i++) {
+            const r = recallResult.relations[i];
+            const sourceId: number | undefined = nameToId[r.source];
+            const targetId: number | undefined = nameToId[r.target];
+            if (sourceId !== undefined && targetId !== undefined) {
+                edgeItems.push({
+                    id: i + 1,
+                    source: sourceId,
+                    target: targetId,
+                    label: r.type,
+                    relation: r.type
+                });
+            }
+        }
+        return edgeItems;
+    }
+
+    /**
+     * 从数据库读取全量图数据，推送给 WebView
+     */
+    private async pushGraphDataToWebView(): Promise<void> {
+        try {
+            const allNodes: GraphNodeItem[] = await this.getAllNodesData();
+            const allEdges: GraphEdgeItem[] = await this.getAllEdgesData();
+
+            if (this.controller) {
+                const jsCode: string =
+                    `window.loadGraphData(${JSON.stringify({ nodes: allNodes, edges: allEdges })});`;
+                this.controller.runJavaScript(jsCode);
+            }
+
+            this.nodeCount = allNodes.length;
+            this.edgeCount = allEdges.length;
+        } catch (err) {
+            Logger.error('pushGraphDataToWebView error: ' + JSON.stringify(err));
+        }
+    }
+
+    build() {
+        Stack() {
+            // WebView 显示 3D 星图
+            Web({ src: $rawfile('graph.html'), controller: this.controller })
+                .javaScriptAccess(true)
+                .width('100%')
+                .height('100%')
+                .zoomAccess(true)
+                .onPageEnd(() => {
+                    this.pushGraphDataToWebView();
+                })
+                .javaScriptProxy({
+                    object: this.bridge,
+                    name: 'nativeBridge',
+                    methodList: ['onNodeClick', 'onSearch'],
+                    asyncMethodList: ['requestGraphData'],
+                    controller: this.controller
+                })
+
+            // 搜索框
+            GraphNodeSearchBar({
+                searchText: this.searchText,
+                onSearchInput: (value: string): void => { this.onSearchInput(value); }
+            })
+
+            // 节点详情浮层
+            if (this.showNodeDetail && this.selectedNodeDetail !== null) {
+                NodeDetailPanel({
+                    detail: this.selectedNodeDetail,
+                    onClose: (): void => { this.closeNodeDetail(); }
+                })
+            }
+        }
+        .width('100%')
+        .height('100%')
+    }
+}
--- a/features/graph/src/main/module.json5
+++ b/features/graph/src/main/module.json5
@ -0,0 +1,12 @@
+{
+  "module": {
+    "name": "graph",
+    "type": "har",
+    "description": "TrulyMEM graph feature module",
+    "deviceTypes": [
+      "phone",
+      "tablet",
+      "2in1"
+    ]
+  }
+}
--- a/products/phone/build/default/intermediates/res/default/resources/rawfile/graph.html
+++ b/products/phone/build/default/intermediates/res/default/resources/rawfile/graph.html
@ -6,13 +6,75 @@
    <title>记忆星图 - TrulyMEM</title>
    <style>
        * { margin: 0; padding: 0; box-sizing: border-box; }
-        body { font-family: 'Courier New', monospace; background: #0a0a1a; overflow: hidden; width: 100vw; height: 100vh; }
+        body { font-family: 'Courier New', monospace; background: #0a0a1a; color: #ffffff; overflow: hidden; width: 100vw; height: 100vh; }
        #canvas-container { width: 100%; height: 100%; position: relative; }
        canvas { display: block; }
+        #stats { position: absolute; top: 20px; left: 20px; background: rgba(10, 10, 26, 0.8); padding: 15px 20px; border-radius: 8px; border: 1px solid rgba(100, 100, 255, 0.3); font-size: 14px; z-index: 100; backdrop-filter: blur(10px); }
+        #stats h3 { margin-bottom: 8px; color: #4488ff; font-size: 16px; }
+        #stats p { margin: 4px 0; color: #aaaacc; }
+        #stats span { color: #ffffff; font-weight: bold; }
+        #node-info { position: absolute; top: 20px; right: 20px; background: rgba(10, 10, 26, 0.9); padding: 15px 20px; border-radius: 8px; border: 1px solid rgba(100, 100, 255, 0.3); font-size: 14px; z-index: 100; display: none; backdrop-filter: blur(10px); max-width: 300px; }
+        #node-info h3 { color: #44ff88; margin-bottom: 8px; font-size: 16px; }
+        #node-info p { margin: 4px 0; color: #aaaacc; }
+        #node-info .label { color: #8888aa; }
+        #loading { position: absolute; top: 50%; left: 50%; transform: translate(-50%, -50%); font-size: 20px; color: #4488ff; z-index: 200; }
+        #nav { position: absolute; bottom: 30px; left: 50%; transform: translateX(-50%); display: flex; gap: 20px; z-index: 100; }
+
+        /* 搜索框 */
+        #search-box { position: absolute; top: 80px; left: 20px; z-index: 100; }
+        #search-input { background: rgba(10, 10, 26, 0.8); border: 1px solid rgba(100, 100, 255, 0.3); color: #ffffff; padding: 8px 12px; border-radius: 6px; font-family: 'Courier New', monospace; font-size: 14px; width: 200px; outline: none; backdrop-filter: blur(10px); }
+        #search-input::placeholder { color: #666688; }
+
+        /* 类型过滤按钮 */
+        #type-filter { position: absolute; top: 120px; left: 20px; display: flex; gap: 8px; z-index: 100; flex-wrap: wrap; }
+        .type-btn { background: rgba(10, 10, 26, 0.8); border: 1px solid rgba(100, 100, 255, 0.3); color: #aaaacc; padding: 6px 12px; border-radius: 6px; cursor: pointer; font-family: 'Courier New', monospace; font-size: 12px; backdrop-filter: blur(10px); transition: all 0.3s; }
+        .type-btn.active { background: rgba(68, 136, 255, 0.3); border-color: #4488ff; color: #ffffff; }
+
+        /* 缩放控制按钮 */
+        #zoom-controls { position: absolute; bottom: 100px; right: 20px; display: flex; flex-direction: column; gap: 10px; z-index: 100; }
+        .zoom-btn { width: 40px; height: 40px; border-radius: 50%; background: rgba(10, 10, 26, 0.8); border: 1px solid rgba(100, 100, 255, 0.3); color: #ffffff; font-size: 20px; cursor: pointer; display: flex; align-items: center; justify-content: center; backdrop-filter: blur(10px); font-family: 'Courier New', monospace; }
+
+        /* 边标签 */
+        #edge-label { position: absolute; display: none; background: rgba(10, 10, 26, 0.9); color: #44ff88; padding: 4px 8px; border-radius: 4px; font-size: 12px; pointer-events: none; z-index: 150; border: 1px solid rgba(68, 255, 136, 0.3); }
+        .nav-btn { background: rgba(10, 10, 26, 0.8); border: 1px solid rgba(100, 100, 255, 0.3); color: #aaaacc; padding: 12px 24px; border-radius: 8px; cursor: pointer; font-family: 'Courier New', monospace; font-size: 14px; backdrop-filter: blur(10px); }
    </style>
 </head>
 <body>
-    <div id="canvas-container"></div>
+    <div id="canvas-container">
+        <div id="loading">正在加载星图数据...</div>
+        <div id="stats">
+            <h3>🌌 记忆星图</h3>
+            <p>节点: <span id="node-count">0</span></p>
+            <p>边: <span id="edge-count">0</span></p>
+            <p>状态: <span id="status">初始化中...</span></p>
+        </div>
+        <div id="node-info">
+            <h3 id="info-name"></h3>
+            <p><span class="label">类型:</span> <span id="info-type"></span></p>
+            <p><span class="label">提及次数:</span> <span id="info-mentions"></span></p>
+            <p><span class="label">连接数:</span> <span id="info-links"></span></p>
+        </div>
+        <div id="nav">
+            <button class="nav-btn active">🌌 星图</button>
+        </div>
+        <div id="search-box">
+            <input type="text" id="search-input" placeholder="搜索节点...">
+        </div>
+        <div id="type-filter">
+            <button class="type-btn active" data-type="全部">全部</button>
+            <button class="type-btn" data-type="Person">Person</button>
+            <button class="type-btn" data-type="Task">Task</button>
+            <button class="type-btn" data-type="AI">AI</button>
+            <button class="type-btn" data-type="Concept">Concept</button>
+            <button class="type-btn" data-type="Object">Object</button>
+        </div>
+        <div id="zoom-controls">
+            <button class="zoom-btn" id="zoom-in">+</button>
+            <button class="zoom-btn" id="zoom-out">-</button>
+            <button class="zoom-btn" id="zoom-reset">R</button>
+        </div>
+        <div id="edge-label"></div>
+    </div>

    <script src="https://cdnjs.cloudflare.com/ajax/libs/three.js/r128/three.min.js"></script>
    <script>
@ -28,9 +90,9 @@
        let activeTypeFilter = '全部';
        let isDragging = false;
        let dragNode = null;
+        let originalPhysicsState = true;
+        let edgeLabelEl = null;
        let nodePositions = {}; // 存储节点位置用于拖拽
-        let nodeCount = 0;
-        let edgeCount = 0;

        const typeColors = { 'person': 0x4488ff, 'task': 0xff8844, 'ai': 0xaa44ff, 'concept': 0x44ff88, 'object': 0xff4444 };
        const defaultColor = 0xcccccc;
@ -59,28 +121,80 @@
            mouse = new THREE.Vector2();
            createStarField();
            createNebula();
+            // 使用 ResizeObserver 监听容器尺寸变化（比 window.resize 更准确）
            initResizeObserver();
            renderer.domElement.addEventListener('mousemove', onMouseMove);
            renderer.domElement.addEventListener('click', onMouseClick);
            window.addEventListener('message', (event) => {
                if (event.data.type === 'graph_data') {
                    window.__graphData = event.data.payload;
-                    loadGraphData(event.data.payload);
+                    loadGraphData();
                }
                if (event.data.type === 'search_nodes') {
                    searchTerm = event.data.query || '';
+                    document.getElementById('search-input').value = searchTerm;
                    applyFilters();
                }
+                if (event.data.type === 'node_detail') {
+                    showNodeDetailPanel(event.data.detail);
+                }
                if (event.data.type === 'highlight_node') {
                    highlightNodeById(event.data.nodeId);
                }
            });
+            // 搜索输入框事件
+            const searchInput = document.getElementById('search-input');
+            searchInput.addEventListener('input', (e) => {
+                searchTerm = e.target.value.toLowerCase();
+                applyFilters();
+                // 通知 ArkTS
+                try {
+                    if (window.nativeBridge && window.nativeBridge.onSearch) {
+                        window.nativeBridge.onSearch(searchTerm);
+                    }
+                } catch(e) {}
+            });
+            // 类型过滤按钮事件
+            document.querySelectorAll('.type-btn').forEach(btn => {
+                btn.addEventListener('click', () => {
+                    document.querySelectorAll('.type-btn').forEach(b => b.classList.remove('active'));
+                    btn.classList.add('active');
+                    activeTypeFilter = btn.dataset.type;
+                    applyFilters();
+                });
+            });
+            // 缩放控制按钮
+            document.getElementById('zoom-in').addEventListener('click', () => {
+                camera.position.multiplyScalar(0.8);
+                controls.update();
+            });
+            document.getElementById('zoom-out').addEventListener('click', () => {
+                camera.position.multiplyScalar(1.2);
+                controls.update();
+            });
+            document.getElementById('zoom-reset').addEventListener('click', () => {
+                if (Object.keys(nodePositions).length > 0) {
+                    const allPositions = Object.values(nodePositions);
+                    let maxDist = 0;
+                    allPositions.forEach(pos => { maxDist = Math.max(maxDist, Math.sqrt(pos.x*pos.x + pos.y*pos.y + pos.z*pos.z)); });
+                    camera.position.set(maxDist * 2.2, maxDist * 1.5, maxDist * 2.2);
+                    controls.target.set(0, 0, 0);
+                    controls.update();
+                }
+            });
+            // 边标签元素
+            edgeLabelEl = document.getElementById('edge-label');
+            // 鼠标移动检测边悬停
+            renderer.domElement.addEventListener('mousemove', onEdgeHoverCheck);
+            // 通知 ArkTS 请求图数据
            try {
                if (window.nativeBridge && window.nativeBridge.requestGraphData) {
                    window.nativeBridge.requestGraphData();
                }
            } catch(e) {}
+            // 触摸事件支持
            initTouchEvents();
+            // 初始化拖拽功能
            initDragFunctionality();
            animate();
        }
@ -146,15 +260,11 @@
            if (data && data.nodes && data.edges) {
                nodes = data.nodes.map(n => ({ id: n.id, name: n.label || n.name, type: n.type, mention_count: n.mentions || 1 }));
                edges = data.edges.map(e => ({ id: e.id, source: e.from || e.source, target: e.to || e.target, relation_type: e.label || e.relation }));
-                nodeCount = nodes.length;
-                edgeCount = edges.length;
+                document.getElementById('node-count').textContent = nodes.length;
+                document.getElementById('edge-count').textContent = edges.length;
+                document.getElementById('status').textContent = '就绪';
                createGraphVisualization();
-                // 通知 ArkTS 统计信息
-                try {
-                    if (window.nativeBridge && window.nativeBridge.onStatsUpdate) {
-                        window.nativeBridge.onStatsUpdate(nodeCount, edgeCount);
-                    }
-                } catch(e) {}
+                document.getElementById('loading').style.display = 'none';
            }
        }

@ -171,7 +281,7 @@
                nodeDegrees[e.target] = (nodeDegrees[e.target] || 0) + 1;
            });
            const positions = {};
-            nodePositions = positions;
+            nodePositions = positions; // 存储供拖拽使用
            const maxDegree = Math.max(...Object.values(nodeDegrees), 1);
            nodes.forEach((node, i) => {
                const angle = (i / nodes.length) * Math.PI * 2;
@ -185,8 +295,8 @@
                        const pos1 = positions[id1], pos2 = positions[id2];
                        const dx = pos1.x - pos2.x, dy = pos1.y - pos2.y, dz = pos1.z - pos2.z;
                        const dist = Math.sqrt(dx*dx + dy*dy + dz*dz) + 0.1;
-                        if (dist < 20) {
-                            const force = 0.3 / (dist * dist);
+                        if (dist < 20) { // 增加排斥距离
+                            const force = 0.3 / (dist * dist); // 增加排斥力系数（平方反比）
                            pos1.x += (dx/dist)*force; pos1.y += (dy/dist)*force; pos1.z += (dz/dist)*force;
                            pos2.x -= (dx/dist)*force; pos2.y -= (dy/dist)*force; pos2.z -= (dz/dist)*force;
                        }
@ -198,7 +308,7 @@
                    const dx = pos2.x - pos1.x, dy = pos2.y - pos1.y, dz = pos2.z - pos1.z;
                    const dist = Math.sqrt(dx*dx + dy*dy + dz*dz) + 0.1;
                    if (dist > 15) {
-                        const force = 0.08;
+                        const force = 0.08; // 增加吸引力
                        pos1.x += (dx/dist)*force; pos1.y += (dy/dist)*force; pos1.z += (dz/dist)*force;
                        pos2.x -= (dx/dist)*force; pos2.y -= (dy/dist)*force; pos2.z -= (dz/dist)*force;
                    }
@ -216,6 +326,7 @@
                sphere.userData = { nodeId: node.id, nodeData: node };
                scene.add(sphere);
                nodeMeshes.push(sphere);
+                // 淡入动画
                fadeInObject(sphere, 500);
            });
            edges.forEach(edge => {
@ -228,6 +339,7 @@
                line.userData = { edgeId: edge.id, edgeData: edge };
                scene.add(line);
                edgeLines.push(line);
+                // 淡入动画
                fadeInLine(line, 500);
            });
            const allPositions = Object.values(positions);
@ -251,9 +363,10 @@
                    if (hoveredNode) hoveredNode.scale.set(1, 1, 1);
                    hoveredNode = node;
                    node.scale.set(1.2, 1.2, 1.2);
+                    showNodeInfo(node.userData.nodeData);
                }
            } else {
-                if (hoveredNode) { hoveredNode.scale.set(1, 1, 1); hoveredNode = null; }
+                if (hoveredNode) { hoveredNode.scale.set(1, 1, 1); hoveredNode = null; hideNodeInfo(); }
            }
        }

@ -265,15 +378,11 @@
                const node = intersects[0].object;
                if (selectedNode === node) {
                    selectedNode = null;
+                    document.getElementById('node-info').style.display = 'none';
                    resetHighlight();
-                    // 通知 ArkTS 取消选中
-                    try {
-                        if (window.nativeBridge && window.nativeBridge.onNodeClick) {
-                            window.nativeBridge.onNodeClick(-1, '');
-                        }
-                    } catch(e) {}
                } else {
                    selectedNode = node;
+                    showNodeInfo(node.userData.nodeData, true);
                    highlightNodeConnections(node.userData.nodeData);
                    // 通知 ArkTS 节点被点击
                    try {
@ -283,54 +392,98 @@
                    } catch(e) {}
                }
            } else {
+                // 点击空白处恢复
                selectedNode = null;
+                document.getElementById('node-info').style.display = 'none';
                resetHighlight();
-                try {
-                    if (window.nativeBridge && window.nativeBridge.onNodeClick) {
-                        window.nativeBridge.onNodeClick(-1, '');
-                    }
-                } catch(e) {}
            }
        }

+        function showNodeInfo(nodeData, isClick = false) {
+            document.getElementById('info-name').textContent = nodeData.name;
+            document.getElementById('info-type').textContent = nodeData.type;
+            document.getElementById('info-mentions').textContent = nodeData.mention_count;
+            const linkCount = edges.filter(e => e.source === nodeData.id || e.target === nodeData.id).length;
+            document.getElementById('info-links').textContent = linkCount;
+            if (isClick) document.getElementById('node-info').style.display = 'block';
+        }
+
+        function hideNodeInfo() { if (!selectedNode) document.getElementById('node-info').style.display = 'none'; }
+
+        // 淡入动画
        function fadeInObject(obj, duration) {
-            const endOpacity = obj.material.transparent ? obj.material.opacity : 1;
-            obj.material.opacity = 0;
+            const startOpacity = 0;
+            const endOpacity = obj.material.opacity !== undefined ? (obj.material.transparent ? obj.material.opacity : 1) : 1;
+            obj.material.opacity = startOpacity;
            obj.material.transparent = true;
            const startTime = Date.now();
            function animateFade() {
                const elapsed = Date.now() - startTime;
                const progress = Math.min(elapsed / duration, 1);
-                obj.material.opacity = progress * endOpacity;
-                if (progress < 1) requestAnimationFrame(animateFade);
-                else obj.material.transparent = endOpacity < 1;
+                obj.material.opacity = startOpacity + (endOpacity - startOpacity) * progress;
+                if (progress < 1) {
+                    requestAnimationFrame(animateFade);
+                } else {
+                    obj.material.transparent = endOpacity < 1;
+                }
            }
            animateFade();
        }

        function fadeInLine(line, duration) {
+            const startOpacity = 0;
            const endOpacity = 0.4;
-            line.material.opacity = 0;
+            line.material.opacity = startOpacity;
            const startTime = Date.now();
            function animateFade() {
                const elapsed = Date.now() - startTime;
                const progress = Math.min(elapsed / duration, 1);
-                line.material.opacity = progress * endOpacity;
-                if (progress < 1) requestAnimationFrame(animateFade);
+                line.material.opacity = startOpacity + (endOpacity - startOpacity) * progress;
+                if (progress < 1) {
+                    requestAnimationFrame(animateFade);
+                }
            }
            animateFade();
        }

+        // 通过节点ID高亮节点（供ArkTS调用）
        function highlightNodeById(nodeId) {
            const mesh = nodeMeshes.find(m => m.userData.nodeData.id === nodeId);
            if (mesh) {
                selectedNode = mesh;
+                showNodeInfo(mesh.userData.nodeData, true);
                highlightNodeConnections(mesh.userData.nodeData);
            }
        }

+        // 显示节点详情面板（供ArkTS调用）
+        function showNodeDetailPanel(detail) {
+            // 更新node-info面板显示详细信息
+            document.getElementById('info-name').textContent = detail.name;
+            document.getElementById('info-type').textContent = detail.type;
+            document.getElementById('info-mentions').textContent = detail.mention_count;
+            document.getElementById('info-links').textContent = detail.connection_count;
+            document.getElementById('node-info').style.display = 'block';
+
+            // 如果有连接信息，添加到面板
+            let detailHtml = `<h3>${detail.name}</h3>`;
+            detailHtml += `<p><span class="label">类型:</span> ${detail.type}</p>`;
+            detailHtml += `<p><span class="label">提及次数:</span> ${detail.mention_count}</p>`;
+            detailHtml += `<p><span class="label">连接数:</span> ${detail.connection_count}</p>`;
+            if (detail.connections && detail.connections.length > 0) {
+                detailHtml += `<p><span class="label">连接关系:</span></p><ul style="margin-left: 15px; font-size: 12px;">`;
+                detail.connections.forEach(conn => {
+                    detailHtml += `<li>${conn.type}: ${conn.target_name}</li>`;
+                });
+                detailHtml += `</ul>`;
+            }
+            document.getElementById('node-info').innerHTML = detailHtml;
+        }
+
+        // 节点拖拽功能
        function initDragFunctionality() {
            let dragStartPos = { x: 0, y: 0 };
+
            renderer.domElement.addEventListener('mousedown', (event) => {
                raycaster.setFromCamera(mouse, camera);
                const intersects = raycaster.intersectObjects(nodeMeshes);
@ -338,14 +491,19 @@
                    isDragging = false;
                    dragNode = intersects[0].object;
                    dragStartPos = { x: event.clientX, y: event.clientY };
+                    originalPhysicsState = true; // 暂停物理模拟
                }
            });
+
            renderer.domElement.addEventListener('mousemove', (event) => {
                if (dragNode) {
                    const dx = event.clientX - dragStartPos.x;
                    const dy = event.clientY - dragStartPos.y;
-                    if (Math.abs(dx) > 3 || Math.abs(dy) > 3) isDragging = true;
+                    if (Math.abs(dx) > 3 || Math.abs(dy) > 3) {
+                        isDragging = true;
+                    }
                    if (isDragging) {
+                        // 将屏幕坐标转换为3D空间
                        const rect = renderer.domElement.getBoundingClientRect();
                        const mouseX = ((event.clientX - rect.left) / rect.width) * 2 - 1;
                        const mouseY = -((event.clientY - rect.top) / rect.height) * 2 + 1;
@ -355,57 +513,79 @@
                        const distance = -camera.position.z / dir.z;
                        const newPos = camera.position.clone().add(dir.multiplyScalar(distance));
                        dragNode.position.copy(newPos);
+                        // 更新存储的位置
                        if (nodePositions[dragNode.userData.nodeId]) {
                            nodePositions[dragNode.userData.nodeId] = { x: newPos.x, y: newPos.y, z: newPos.z };
                        }
                    }
                }
            });
+
            renderer.domElement.addEventListener('mouseup', () => {
-                if (dragNode) { dragNode = null; isDragging = false; }
+                if (dragNode) {
+                    // 恢复物理模拟
+                    dragNode = null;
+                    isDragging = false;
+                }
            });
        }

+        // Touch事件支持
        function initTouchEvents() {
-            let touchStartPos = { x: 0, y: 0 };
+            let touchStart = null;
            let touchStartDistance = 0;
+            let touchStartPos = { x: 0, y: 0 };
            let isTouchDrag = false;
+
            renderer.domElement.addEventListener('touchstart', (event) => {
                event.preventDefault();
                if (event.touches.length === 1) {
                    const touch = event.touches[0];
                    touchStartPos = { x: touch.clientX, y: touch.clientY };
                    isTouchDrag = false;
+
+                    // 模拟鼠标事件用于射线检测
                    const rect = renderer.domElement.getBoundingClientRect();
                    mouse.x = ((touch.clientX - rect.left) / rect.width) * 2 - 1;
                    mouse.y = -((touch.clientY - rect.top) / rect.height) * 2 + 1;
+
                    raycaster.setFromCamera(mouse, camera);
                    const intersects = raycaster.intersectObjects(nodeMeshes);
                    if (intersects.length > 0) {
                        const node = intersects[0].object;
                        if (selectedNode === node) {
                            selectedNode = null;
+                            document.getElementById('node-info').style.display = 'none';
                            resetHighlight();
-                            try { if (window.nativeBridge && window.nativeBridge.onNodeClick) window.nativeBridge.onNodeClick(-1, ''); } catch(e) {}
                        } else {
                            selectedNode = node;
+                            showNodeInfo(node.userData.nodeData, true);
                            highlightNodeConnections(node.userData.nodeData);
-                            try { if (window.nativeBridge && window.nativeBridge.onNodeClick) window.nativeBridge.onNodeClick(node.userData.nodeData.id, node.userData.nodeData.name); } catch(e) {}
+                            try {
+                                if (window.nativeBridge && window.nativeBridge.onNodeClick) {
+                                    window.nativeBridge.onNodeClick(node.userData.nodeData.id, node.userData.nodeData.name);
+                                }
+                            } catch(e) {}
                        }
                    }
                } else if (event.touches.length === 2) {
+                    // 双指缩放
                    const dx = event.touches[0].clientX - event.touches[1].clientX;
                    const dy = event.touches[0].clientY - event.touches[1].clientY;
                    touchStartDistance = Math.sqrt(dx*dx + dy*dy);
                }
            }, { passive: false });
+
            renderer.domElement.addEventListener('touchmove', (event) => {
                event.preventDefault();
                if (event.touches.length === 1 && controls) {
                    const touch = event.touches[0];
                    const dx = touch.clientX - touchStartPos.x;
                    const dy = touch.clientY - touchStartPos.y;
-                    if (Math.abs(dx) > 5 || Math.abs(dy) > 5) isTouchDrag = true;
+                    if (Math.abs(dx) > 5 || Math.abs(dy) > 5) {
+                        isTouchDrag = true;
+                    }
+                    // 模拟OrbitControls的鼠标移动
                    if (isTouchDrag) {
                        const rotateSpeed = 0.005;
                        controls.rotateLeft(-dx * rotateSpeed);
@ -414,6 +594,7 @@
                        touchStartPos = { x: touch.clientX, y: touch.clientY };
                    }
                } else if (event.touches.length === 2 && controls) {
+                    // 双指缩放
                    const dx = event.touches[0].clientX - event.touches[1].clientX;
                    const dy = event.touches[0].clientY - event.touches[1].clientY;
                    const distance = Math.sqrt(dx*dx + dy*dy);
@ -423,16 +604,21 @@
                    touchStartDistance = distance;
                }
            }, { passive: false });
+
            renderer.domElement.addEventListener('touchend', (event) => {
-                if (event.touches.length === 0) isTouchDrag = false;
+                if (event.touches.length === 0) {
+                    isTouchDrag = false;
+                }
            }, { passive: false });
        }

+        // 搜索和类型过滤
        function applyFilters() {
            nodeMeshes.forEach(mesh => {
                const nodeData = mesh.userData.nodeData;
                const nameMatch = nodeData.name.toLowerCase().includes(searchTerm);
                const typeMatch = activeTypeFilter === '全部' || nodeData.type === activeTypeFilter;
+
                if (nameMatch && typeMatch) {
                    mesh.material.transparent = false;
                    mesh.material.opacity = 1;
@ -443,12 +629,14 @@
                    mesh.scale.setScalar(0.8);
                }
            });
+            // 同时过滤边
            edgeLines.forEach(line => {
                const edgeData = line.userData.edgeData;
                const sourceNode = nodes.find(n => n.id === edgeData.source);
                const targetNode = nodes.find(n => n.id === edgeData.target);
                const sourceMatch = sourceNode && sourceNode.name.toLowerCase().includes(searchTerm) && (activeTypeFilter === '全部' || sourceNode.type === activeTypeFilter);
                const targetMatch = targetNode && targetNode.name.toLowerCase().includes(searchTerm) && (activeTypeFilter === '全部' || targetNode.type === activeTypeFilter);
+
                if ((sourceMatch || targetMatch) && searchTerm === '' && activeTypeFilter === '全部') {
                    line.material.transparent = true;
                    line.material.opacity = 0.4;
@ -462,9 +650,12 @@
            });
        }

+        // 连接高亮
        function highlightNodeConnections(nodeData) {
            const connectedNodeIds = new Set();
            const connectedEdgeIds = new Set();
+
+            // 找出所有连接的节点和边
            edges.forEach(edge => {
                if (edge.source === nodeData.id || edge.target === nodeData.id) {
                    connectedNodeIds.add(edge.source);
@ -472,30 +663,37 @@
                    connectedEdgeIds.add(edge.id);
                }
            });
+
            nodeMeshes.forEach(mesh => {
                const meshNodeId = mesh.userData.nodeData.id;
                if (meshNodeId === nodeData.id) {
+                    // 选中的节点
                    mesh.material.emissiveIntensity = 1.0;
                    mesh.scale.setScalar(1.5);
                } else if (connectedNodeIds.has(meshNodeId)) {
+                    // 直接连接的节点
                    mesh.material.emissiveIntensity = 0.8;
                    mesh.scale.setScalar(1.3);
                    mesh.material.transparent = false;
                    mesh.material.opacity = 1;
                } else {
+                    // 无关系的节点
                    mesh.material.transparent = true;
                    mesh.material.opacity = 0.15;
                    mesh.material.emissiveIntensity = 0.2;
                    mesh.scale.setScalar(0.9);
                }
            });
+
            edgeLines.forEach(line => {
                if (connectedEdgeIds.has(line.userData.edgeData.id)) {
                    line.material.transparent = true;
                    line.material.opacity = 0.8;
+                    line.material.linewidth = 2;
                } else {
                    line.material.transparent = true;
                    line.material.opacity = 0.1;
+                    line.material.linewidth = 1;
                }
            });
        }
@ -505,6 +703,7 @@
                const nodeData = mesh.userData.nodeData;
                const nameMatch = nodeData.name.toLowerCase().includes(searchTerm);
                const typeMatch = activeTypeFilter === '全部' || nodeData.type === activeTypeFilter;
+
                mesh.material.emissiveIntensity = 0.5 + Math.min(nodeData.mention_count * 0.05, 0.3);
                if (nameMatch && typeMatch) {
                    mesh.material.transparent = false;
@ -519,9 +718,32 @@
            edgeLines.forEach(line => {
                line.material.transparent = true;
                line.material.opacity = 0.4;
+                line.material.linewidth = 1;
            });
        }

+        // 边标签显示
+        function onEdgeHoverCheck(event) {
+            const rect = renderer.domElement.getBoundingClientRect();
+            mouse.x = ((event.clientX - rect.left) / rect.width) * 2 - 1;
+            mouse.y = -((event.clientY - rect.top) / rect.height) * 2 + 1;
+            raycaster.setFromCamera(mouse, camera);
+
+            // 检查边悬停
+            const edgeIntersects = raycaster.intersectObjects(edgeLines);
+            if (edgeIntersects.length > 0) {
+                const edge = edgeIntersects[0].object;
+                const edgeData = edge.userData.edgeData;
+                edgeLabelEl.textContent = edgeData.relation_type || '关系';
+                edgeLabelEl.style.display = 'block';
+                edgeLabelEl.style.left = (event.clientX - rect.left + 10) + 'px';
+                edgeLabelEl.style.top = (event.clientY - rect.top - 10) + 'px';
+            } else {
+                edgeLabelEl.style.display = 'none';
+            }
+        }
+
+        // ========= ResizeObserver 响应式适配 =========
        let containerObserver = null;
        function initResizeObserver() {
            const container = document.getElementById('canvas-container');
@ -529,7 +751,9 @@
            containerObserver = new ResizeObserver((entries) => {
                for (const entry of entries) {
                    const { width, height } = entry.contentRect;
-                    if (width > 0 && height > 0) onContainerResize(width, height);
+                    if (width > 0 && height > 0) {
+                        onContainerResize(width, height);
+                    }
                }
            });
            containerObserver.observe(container);
@ -541,10 +765,55 @@
            camera.aspect = aspect;
            camera.updateProjectionMatrix();
            renderer.setSize(width, height);
+
+            // 窄屏(<500px)时自动调整UI元素尺寸和位置
+            const isNarrow = width < 500;
+            const stats = document.getElementById('stats');
+            const nodeInfo = document.getElementById('node-info');
+            const searchBox = document.getElementById('search-box');
+            const typeFilter = document.getElementById('type-filter');
+            const searchInput = document.getElementById('search-input');
+
+            if (isNarrow) {
+                if (stats) {
+                    stats.style.fontSize = '11px';
+                    stats.style.padding = '8px 12px';
+                    stats.style.top = '6px';
+                    stats.style.left = '6px';
+                }
+                if (nodeInfo) {
+                    nodeInfo.style.fontSize = '11px';
+                    nodeInfo.style.padding = '8px 12px';
+                    nodeInfo.style.maxWidth = '180px';
+                    nodeInfo.style.top = '6px';
+                    nodeInfo.style.right = '6px';
+                }
+                if (searchBox) { searchBox.style.top = '70px'; searchBox.style.left = '6px'; }
+                if (searchInput) { searchInput.style.width = '140px'; searchInput.style.fontSize = '12px'; }
+                if (typeFilter) { typeFilter.style.top = '108px'; typeFilter.style.left = '6px'; }
+            } else {
+                if (stats) {
+                    stats.style.fontSize = '14px';
+                    stats.style.padding = '15px 20px';
+                    stats.style.top = '20px';
+                    stats.style.left = '20px';
+                }
+                if (nodeInfo) {
+                    nodeInfo.style.fontSize = '14px';
+                    nodeInfo.style.padding = '15px 20px';
+                    nodeInfo.style.maxWidth = '300px';
+                    nodeInfo.style.top = '20px';
+                    nodeInfo.style.right = '20px';
+                }
+                if (searchBox) { searchBox.style.top = '80px'; searchBox.style.left = '20px'; }
+                if (searchInput) { searchInput.style.width = '200px'; searchInput.style.fontSize = '14px'; }
+                if (typeFilter) { typeFilter.style.top = '120px'; typeFilter.style.left = '20px'; }
+            }
        }

        function animate() {
            animationId = requestAnimationFrame(animate);
+            const time = Date.now() * 0.001;
            highlightPulse = (highlightPulse + 0.02) % (Math.PI * 2);
            controls.update();
            if (starField) starField.rotation.y += 0.0001;
@ -555,4 +824,4 @@
        init();
    </script>
 </body>
-</html>
+</html>
--- a/features/settings/BuildProfile.ets
+++ b/features/settings/BuildProfile.ets
@ -0,0 +1,17 @@
+/**
+ * Use these variables when you tailor your ArkTS code. They must be of the const type.
+ */
+export const HAR_VERSION = '1.0.0';
+export const BUILD_MODE_NAME = 'debug';
+export const DEBUG = true;
+export const TARGET_NAME = 'default';
+
+/**
+ * BuildProfile Class is used only for compatibility purposes.
+ */
+export default class BuildProfile { 
+	static readonly HAR_VERSION = HAR_VERSION;
+	static readonly BUILD_MODE_NAME = BUILD_MODE_NAME;
+	static readonly DEBUG = DEBUG;
+	static readonly TARGET_NAME = TARGET_NAME;
+}
--- a/features/settings/Index.ets
+++ b/features/settings/Index.ets
@ -0,0 +1 @@
+export { SettingsPage } from './src/main/ets/pages/SettingsPage';
--- a/features/settings/build-profile.json5
+++ b/features/settings/build-profile.json5
@ -0,0 +1,10 @@
+{
+  "apiType": "stageMode",
+  "buildOption": {
+  },
+  "targets": [
+    {
+      "name": "default"
+    }
+  ]
+}
--- a/features/settings/build/default/intermediates/merge_profile/default/module.json
+++ b/features/settings/build/default/intermediates/merge_profile/default/module.json
@ -1,30 +0,0 @@
-{
-	"app": {
-		"bundleName": "com.trulymem.app",
-		"debug": true,
-		"versionCode": 1000001,
-		"versionName": "1.0.0",
-		"minAPIVersion": 60100023,
-		"targetAPIVersion": 60100023,
-		"apiReleaseType": "Release",
-		"targetMinorAPIVersion": 0,
-		"targetPatchAPIVersion": 0,
-		"compileSdkVersion": "6.1.0.105",
-		"compileSdkType": "HarmonyOS",
-		"appEnvironments": [],
-		"bundleType": "app",
-		"buildMode": "debug"
-	},
-	"module": {
-		"name": "settings",
-		"type": "har",
-		"description": "TrulyMEM settings feature module",
-		"deviceTypes": [
-			"phone",
-			"tablet",
-			"2in1"
-		],
-		"packageName": "@ohos/settings",
-		"installationFree": false
-	}
-}
--- a/features/settings/hvigorfile.ts
+++ b/features/settings/hvigorfile.ts
@ -0,0 +1,6 @@
+import { harTasks } from '@ohos/hvigor-ohos-plugin';
+
+export default {
+    system: harTasks,
+    plugins: []
+};
--- a/features/settings/oh-package-lock.json5
+++ b/features/settings/oh-package-lock.json5
@ -0,0 +1,19 @@
+{
+  "meta": {
+    "stableOrder": true,
+    "enableUnifiedLockfile": false
+  },
+  "lockfileVersion": 3,
+  "ATTENTION": "THIS IS AN AUTOGENERATED FILE. DO NOT EDIT THIS FILE DIRECTLY.",
+  "specifiers": {
+    "@ohos/common@../../common": "@ohos/common@../../common"
+  },
+  "packages": {
+    "@ohos/common@../../common": {
+      "name": "@ohos/common",
+      "version": "1.0.0",
+      "resolved": "../../common",
+      "registryType": "local"
+    }
+  }
+}
--- a/features/settings/oh-package.json5
+++ b/features/settings/oh-package.json5
@ -0,0 +1,11 @@
+{
+  "name": "@ohos/settings",
+  "version": "1.0.0",
+  "description": "TrulyMEM settings feature module",
+  "main": "Index.ets",
+  "author": "",
+  "license": "",
+  "dependencies": {
+    "@ohos/common": "file:../../common"
+  }
+}
--- a/features/settings/oh_modules/@ohos/common
+++ b/features/settings/oh_modules/@ohos/common
@ -0,0 +1 @@
+../../../../common
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
root	e779c16595	refactor: 工程标准化 — 组件提取、废弃API清理、console清零 - ChatPage/GraphPage/SettingsPage 拆分为独立子组件 - MainPage 接入 ImmersiveTabNavigation 玻璃拟态导航 - EntryAbility console.* 全替换为 defaultLogger - SplashPage replaceUrl 废弃 API 迁移到 UIContext Router - common/Index.ets 双通道导出 Logger/defaultLogger - 删除 dead code (entry/, commonbusiness/, trulymem-core/) - 全工程 0 console.*，BUILD SUCCESSFUL	2026-05-01 16:05:22 +08:00
root	5e4fb8222a	refactor: 工程规范化，唯一入口为 products/phone - entry 模块改为 har（静态共享包），移除 abilities 和入口能力 - entry/build-profile.json5 简化，删除 release 混淆配置 - 根 build-profile.json5 移除 entry 模块，phone 为唯一入口 - 保留 entry 的 requestPermissions 作为依赖参考	2026-05-01 09:22:54 +08:00
root	5ae2143822	refactor: 改用原生 Tabs 导航，清理旧 entry 目录代码 - MainPage 改用 ArkUI Tabs + BottomTabBarStyle 原生导航 - 移除 ImmersiveTabNavigation 自定义组件依赖 - 清理旧 entry/src/main/ets/ 下的已迁移文件 - 更新 common/Index.ets 移除 ImmersiveTabNavigation 导出	2026-05-01 09:21:39 +08:00
root	ab3d910387	refactor: 应用参考工程代码模式，优化启动流程和架构 - 新增 SplashPage：启动页，2秒后跳转到 Index - 新增 MainPage：独立主页组件，含 Tab 导航 + 宽屏/窄屏自适应 - 简化 Index.ets：纯路由入口，初始化数据库，加载 MainPage - EntryAbility：加载 SplashPage 而非直接加载 Index - 新增 PageContext：NavPathStack 路由管理（参考工程模式） - 新增 BreakpointSystem：响应式断点系统 xs/sm/md/lg/xl - 新增 BaseViewModel：ViewModel 基类，含 attach/detach/dispose 生命周期 - 更新 common/Index.ets 导出新模块 - 更新 main_pages.json 注册 SplashPage 和 MainPage	2026-05-01 09:15:36 +08:00
root	5916d2749c	refactor: 按 sample_in_harmonyos 多层模块化架构重构工程结构 - 新增 common/ 公共模块：模型、服务、组件、常量 - 新增 features/ 功能模块：graph(星图)、chat(聊天)、settings(设置)、commonbusiness - 新增 products/phone 产品入口层：EntryAbility + Index + MainPage - Index.ets 简化为纯路由入口，TabNavigation 移至 MainPage 统一管理 - 更新 build-profile.json5 注册所有新模块 - 迁移原始 entry/ 代码到分层架构，保持功能完整	2026-05-01 08:35:07 +08:00
root	67c6b38e50	fix: resolve ArkTS compilation errors - GraphDatabase.ets: - Export TimeRangeParams interface - Add graph(), snapshot(), queryArchived() methods - Add SnapshotData interface - Add NameCacheEntry interface, remove duplicates - Fix Context import to use named import from @ohos.abilityAccessCtrl - GraphMemoryService.ets: - Import TimeRangeParams from GraphDatabase - Import RecordResult from GraphDatabase - Add graph(), snapshot(), queryArchived() delegate methods - Add timeRange optional param to MemoryRecallParams - AIAgentService.ets: - Fix Context import: named import from @ohos.abilityAccessCtrl - Import TimeRangeParams from GraphDatabase instead of GraphMemoryService - Add dryRun, keyword, attribute to ToolPropertiesDefinition - Index.ets: - Fix displayCallback type to Callback<number> - Add display import	2026-04-30 17:41:02 +08:00
root	1a4a8dce98	feat: 移植6个缺失工具定义和handler到AIAgentService - memory_introspect — 查看记忆状态统计 - memory_archive — 归档旧记忆 - memory_cleanup — 清理已删除数据 - memory_query_archived — 查询已归档记忆 - context_rewrite — 压缩工具调用上下文 - persona_remove — 删除单条人设属性同时在GraphMemoryService添加： - queryArchived() 方法 - personaRemove() 方法更新系统提示词，加入工具调用规则和context_rewrite使用说明	2026-04-30 16:44:27 +08:00
root	c8571f9d5f	perf: removeOrphanNodes O(N²) → O(N) 批量查询优化 - 改为先批量查询所有 active 关系的节点 ID 集合 - 再遍历所有节点找出孤儿节点 - 从 O(N×2) 查询降到 O(1) 批量 + O(N) 遍历 - 显著提升 cleanup / purge 性能	2026-04-30 15:16:19 +08:00
root	dd0ab4dcbe	perf: recall BFS N+1 优化 + timeRange 过滤支持 - GraphDatabase.recall() 添加 timeRange.days 参数，计算 minDateBucket 过滤 - getRelationsForNodes() 批量预加载节点名称缓存，减少 N+1 查询 - 关系查询传入 minDateBucket，支持按时间范围过滤关系 - BFS 扩展节点时优先从缓存获取，避免重复查询数据库	2026-04-30 15:09:53 +08:00
root	ba44d50d7a	fix: AIAgentService getContext(this) 改为外部传入，Index 添加 aboutToDisappear 释放 display 监听 - AIAgentService: 添加 appContext 字段，构造函数接收 Context 参数 - ChatPage: 创建 AIAgentService 时传入 getContext(this) - Index: 添加 displayCallback 字段和 aboutToDisappear 释放监听	2026-04-30 15:02:00 +08:00
root	b9a2cba2a7	fix: 编译错误修复 - ArkTS语法兼容(去交点类型/展开/内联对象类型/duplicate code) + pinchGesture/alignContent移除	2026-04-30 10:35:54 +08:00
root	c060a59710	feat: 星图响应式布局适配 - ResizeObserver + 窄屏55/45上下分/宽屏左右分 (@ohos.display)	2026-04-30 10:29:51 +08:00
root	d594099f02	feat: harmonyos星图全面适配交互功能-搜索/过滤/连接高亮/拖拽/触摸/节点详情	2026-04-30 10:24:06 +08:00
root	45c098d10b	chore: update harmonyos source files	2026-04-29 23:30:02 +08:00
root	ce7d3b532f	chore: remove remaining __pycache__ files, add .gitignore	2026-04-28 17:09:37 +08:00
root	4bfa4673be	chore: clean harmonyos branch - keep only DevEco source files (57 files)	2026-04-28 17:09:15 +08:00
root	e6dc56b985	chore: remove HarmonyOS build cache files from git tracking (69 files)	2026-04-28 17:06:09 +08:00
root	11ee329943	refactor: move HarmonyOS project files from harmony/ to branch root	2026-04-28 16:52:05 +08:00
root	4c32dbb949	chore: remove legacy .gitignore, only keep HarmonyOS project files	2026-04-28 16:46:09 +08:00
root	5c59cd2ceb	refactor: clean harmonyos branch to only contain HarmonyOS project in harmony/	2026-04-28 16:34:34 +08:00
root	e83151bd1e	refactor: move harmonyos project to harmony/ directory	2026-04-28 16:32:04 +08:00
root	2a42183aaa	feat: 合并星图和聊天页面为 MainPage，响应式 GridRow 布局 - 新建 MainPage.ets，用 GridRow 响应式布局组合星图+聊天 - Index.ets 改为 2 Tab（TrulyMEM + 设置） - GraphPage/ChatPage 接收 @Prop db 从父组件传入 - 恢复 README.md / README_EN.md 的特别鸣谢部分	2026-04-28 13:53:06 +08:00
root	7c2441682c	feat: 纯血鸿蒙 ArkTS 工程（harmonyos 分支） - 新增 harmonyos/ 目录：完整 ArkTS 项目 - GraphDatabase.ets：@ohos.data.relationalStore 图数据库 - Index.ets：Tabs 导航（星图\|聊天\|设置） - GraphPage.ets：WebView 星图（Three.js 3D） - ChatPage.ets：聊天 + HTTP AI API 调用 - SettingsPage.ets：@ohos.data.preferences 配置 - graph.html：Three.js 力导向星图可视化 - 删除：ui/ TUI 代码、core/migrate.py 多用户迁移、trulymem_entry.py - 配置：build-profile.json5, module.json5, permissions	2026-04-28 12:38:21 +08:00
root	14bbcfde3d	chore: 清理误提交的构建产物，添加 .gitignore 规则	2026-04-28 12:34:33 +08:00
root	be6555cca1	chore: 删除 TUI/多用户/CLI 代码，准备鸿蒙工程	2026-04-28 12:34:26 +08:00
root	c4f7b5a645	feat: 入口添加 --web 参数，二进制可直接启动 Web headless 服务	2026-04-28 11:24:15 +08:00
root	1a971fee73	fix: 修复 PyInstaller 构建 - 修复 spec 文件中 project_root 路径计算（需向上两级） - 修复 datas 目标路径：PyInstaller data tuple 的第二个元素必须是目录路径，而非文件路径 - 移除 PYZ(block_cipher) 参数避免 TypeError	2026-04-28 11:05:55 +08:00
root	4770b3b990	refactor: Web 服务内嵌主进程 + 文档移到 docs/ 二级目录 1. Web 服务改为 threading 模式： - 新增 run_web_server(port) / stop_web_server() 函数 - TUI 通过 from web_api import run_web_server 直接调用 - 使用 werkzeug make_server 实现线程内优雅启动/停止 2. 构建脚本精简为单二进制： - 移除 trulymem-web 独立打包步骤 - 添加 werkzeug 到 hidden-imports 3. 详细启动文档移到 docs/zh/quick_start.md 4. README 精简为概述，指向 docs/ 目录	2026-04-28 10:50:24 +08:00
				`@ -0,0 +1 @@`
				`{"SECRET_KEY": "3a38a0f46a76673154b491a7b061c38f2f6a55489078ae7b5d34e94e75fc0534"}`
				`@ -0,0 +1 @@`
				`export { ChatPage } from './src/main/ets/pages/ChatPage';`
				`@ -0,0 +1 @@`
				`/home/program/TrulyMEM-TrueHumanMEM/features/chat`
				`@ -0,0 +1 @@`
				`export { GraphPage } from './src/main/ets/pages/GraphPage';`
				`@ -0,0 +1 @@`
				`export { SettingsPage } from './src/main/ets/pages/SettingsPage';`