chore(gui): 版本号 3.0.0（与内核同版本）

改名为 breaking：go module 路径、二进制名、`/etc` 配置目录、`LLMSPROXY_*`
环境变量全部变更，已部署实例必须跑 `packaging/migrate-to-modelrouter.sh`。
因此内核与桌面端一起升到 3.0.0，而不是继续在 v2.0.x 上打 patch。

`Makefile` 的 VERSION 来自 `git describe`，无需另改。

HANDOFF_2026-10-19.md

@@ -0,0 +1,497 @@
+# ModelRouter / llmsproxy 交接文档
+
+更新时间：2026-10-19
+工作目录：`/home/program/llmsproxy`
+当前分支：`release/v2.0.x`
+
+## 1. 当前 Git 状态
+
+最近提交：
+
+- `c9f3453 fix(codearts): 识别 InferHub 未注册模型`
+- `12bfd53 fix: 收口渠道源迁移与测试页合并`
+- `25970a2 refactor(scheduler)!: 移除 priority 废弃字段 + 一通道一源 + discover 复用凭据`
+
+`c9f3453` 已推送到 `origin/release/v2.0.x`。
+
+注意：当前工作树在最后一轮添加 provider 回归测试时出现了一个**未完成且可能无法编译的测试编辑**。最后一次工具报错：
+
+```text
+internal/provider/provider_test.go:99: missing ',' in argument list
+```
+
+下一位 agent 接手后的第一步必须是：
+
+```bash
+cd /home/program/llmsproxy
+git status --short
+gofmt -w internal/provider/provider_test.go 2>&1 || true
+go test -tags luajit ./internal/provider/ 2>&1
+```
+
+如果测试文件处于半编辑状态，先检查 `TestProviderChatStreamNormalizesStopAfterToolCall`，修正 Go 字符串转义或删除该未完成测试后再继续。不要直接部署当前工作树。
+
+## 2. 生产服务状态
+
+生产服务：`llmsproxy.service`
+
+- 当前服务 active
+- 监听：`0.0.0.0:8081`
+- 二进制已通过 `./deploy.sh` 原子替换并重启
+- 最近一次部署健康检查通过
+- 当前生产插件 SHA256 前缀曾核对为：`1bf26cbb3abcb107`
+- 当前生产二进制在最近部署后已更新
+
+不要把密钥写入交接记录。生产 admin key 保存在本机运维环境中，不应复制到聊天、文档或提交。
+
+## 3. Channel 源当前状态
+
+生产源已经从旧的“一账号一源”收口成“一渠道一源”：
+
+```text
+channels-trae
+channels-minimax
+channels-lobsterai
+channels-buddy
+channels-codearts
+channels-zcode
+```
+
+旧形态源已经清理：
+
+```text
+channels-<provider>-<account_id>
+```
+
+此前旧源和新源并存，造成同一渠道同一模型出现多份，只能靠账号 hex 区分。
+
+根因已修复：内核 `list_sources` 返回裸数组，但旧插件代码按 `{sources = [...]}` 信封读取，导致迁移清理循环永远为空。当前 `channels.lua` 已兼容：
+
+- 裸数组
+- `{sources = [...]}`
+- `{json = {sources = [...]}}`
+
+相关代码：
+
+```text
+internal/lua/plugins/channels.lua: syncProviderSource
+internal/core/channels_legacy_source_cleanup_test.go
+```
+
+当前生产源列表已实测只有 6 个新形态源。
+
+## 4. CodeArts 当前状态
+
+### 已确认可用的部分
+
+CodeArts 使用：
+
+```text
+adapter = codearts
+base_url = https://snap-access.cn-north-4.myhuaweicloud.com
+```
+
+CodeArts 的 chat 请求需要完整 AK/SK/security token 凭据，由插件适配器签名，不是裸 Bearer token。
+
+真实流式工具调用测试已经成功：
+
+请求形状：
+
+```json
+{
+  "model": "channels-codearts:openpangu-2.0-flash",
+  "messages": [{"role": "user", "content": "Use the weather tool for Beijing."}],
+  "tools": [{
+    "type": "function",
+    "function": {
+      "name": "get_weather",
+      "description": "Get weather",
+      "parameters": {
+        "type": "object",
+        "properties": {"city": {"type": "string"}},
+        "required": ["city"]
+      }
+    }
+  }],
+  "tool_choice": "auto",
+  "stream": true,
+  "max_tokens": 64
+}
+```
+
+结果：
+
+- HTTP 200
+- `Content-Type: text/event-stream`
+- 返回结构化 `tool_calls`
+- 函数名 `get_weather`
+- 参数最终为 `{"city":"Beijing"}`
+- `finish_reason = tool_calls`
+
+因此 CodeArts 的流式工具调用链路当前是通的。
+
+### CodeArts 模型刷新
+
+生产调用：
+
+```http
+POST /api/plugins/http/channels/model/refresh
+{"platform":"codearts"}
+```
+
+实测返回：
+
+```json
+{
+  "ok": true,
+  "results": [
+    {"platform": "codearts", "ok": true, "count": 11}
+  ]
+}
+```
+
+CodeArts 模型目录来自两个签名端点并合并：
+
+1. `opengw.developer.huaweicloud.com/api/v1/gateway/config`
+   - `result.models`
+2. `snap-access.cn-north-4.myhuaweicloud.com/v1/model/builtin`
+   - `builtinModels`
+
+相关解析：
+
+```text
+internal/lua/plugins/channels.lua: parseRemoteModels
+internal/lua/plugins/channels.lua: refreshRemoteModels
+internal/core/channels_codearts_catalog_test.go
+```
+
+### 已确认失效并禁用的 CodeArts 模型
+
+下面 4 个模型曾收到明确的上游 404：
+
+```text
+CodeArts-Pangu
+deepseek-v4-flash-0731
+deepseek-v4-pro-0813
+glm-5.3-flash
+```
+
+生产上已通过 `channels/model/set_disabled` 禁用。它们仍可能在 `model/list` 中出现，但 `disabled=true`，不会作为可用模型路由。
+
+新增的死模型判据：
+
+```lua
+model is not registered
+```
+
+对应提交：`c9f3453`。
+
+注意：当前插件的 dead-model 机制只在收到真实请求失败后记录模型；刷新目录不会自动根据“远端目录没有该模型”删除模型，这是刻意设计，避免误删远端目录不完整的平台模型。
+
+CodeArts 当前仍有一些模型返回 429/TPM 限流，例如：
+
+```text
+deepseek-v4-flash
+deepseek-v4.1-flash
+```
+
+429 不是模型失效，不能禁用。
+
+## 5. 其他 Channel 实测结果
+
+### Trae
+
+真实流式 + tools 请求：
+
+```text
+channels-trae:DeepSeek-V4-Flash
+```
+
+结果：
+
+- HTTP 200
+- SSE 正常返回
+- 结构化 tool call 返回
+- 但上游最后的结束原因可能是 `stop`，即使前面已经出现 tool-call fragments
+
+这会让 Agent 把工具调用误判成普通文本完成。
+
+当前准备中的修复在统一 provider 流式层：
+
+```text
+internal/provider/provider.go
+```
+
+逻辑是：只要本次流出现真实 tool-call 片段，后续 `FinishReason == "stop"` 就归一成 `tool_calls`。
+
+另外，Trae 的非流式接口偶尔返回 SSE。当前 `trae.lua` 对非流式 SSE 只取最后一条 data，不能完整合并多条工具调用/推理片段。这是后续仍需处理的协议问题，但不能用非流式手工请求结果推断 Pi 的真实请求模式。
+
+### Buddy
+
+真实流式 + tools 请求：
+
+```text
+channels-buddy:glm-5.3-flash
+```
+
+结果：
+
+- HTTP 200
+- SSE 正常返回
+- tool-call 参数分片可见
+- 最终有 `finish_reason = tool_calls`
+- 但某些上游 chunk 的 `tool_calls` 可能是空对象 `{}`
+
+空对象不是合法的 OpenAI tool-call 数组，部分 Agent 客户端会拒绝。
+
+当前准备中的修复在：
+
+```text
+internal/lua/adapters/buddy.lua
+```
+
+只在 `type(delta.tool_calls) == "table" and #delta.tool_calls > 0` 时输出 `tool_calls`，避免 Lua 空表被编码为 `{}`。
+
+Buddy 的非流式错误：
+
+```text
+11101: Non-stream chat request is currently not supported
+```
+
+之前出现的该错误来自手工 `curl` 测试中显式设置 `stream:false`，不能归因到 Pi。真实 `stream:true + tools` 已经证明 Buddy 流式工具调用可以工作。
+
+### MiniMax
+
+生产日志反复出现：
+
+```text
+adapter "anthropic": status=401 body={"error":"invalid access token"}
+provider channels-minimax: api error 401
+```
+
+这是凭据失效问题，不是 tools 形状问题。
+
+当前源：
+
+```text
+channels-minimax
+adapter = anthropic
+endpoint = /mavis/api/v1/llm/v1/messages
+```
+
+需要用户重新登录/提供有效凭据。不要擅自刷新一次性 refresh token，之前已经发生过 refresh token 被消耗后无法再次使用的事故。
+
+## 6. 当前未提交的代码改动
+
+最后一轮已经改动但尚未完成提交：
+
+```text
+internal/provider/provider.go
+internal/lua/adapters/trae.lua
+internal/lua/adapters/buddy.lua
+internal/provider/provider_test.go  # 可能有半编辑、需先修复
+```
+
+已完成的意图：
+
+1. provider 流式层记录 `sawToolCalls`
+2. 看到真实 tool-call 后，把后续 `stop` 归一为 `tool_calls`
+3. Trae 不输出空工具调用表
+4. Buddy 不输出空工具调用表
+5. provider 回归测试模拟“tool fragment + stop”场景
+
+但 provider 测试最后一次编辑工具报错：
+
+```text
+internal/provider/provider_test.go:99: missing ',' in argument list
+```
+
+因此必须先修测试文件，再提交。
+
+## 7. 推荐接手顺序
+
+### 第一步：修复工作树
+
+```bash
+cd /home/program/llmsproxy
+git status --short
+sed -n '1,150p' internal/provider/provider_test.go
+gofmt -w internal/provider/provider_test.go 2>&1 || true
+go test -tags luajit ./internal/provider/ 2>&1
+```
+
+若测试函数字符串转义难以修复，先删除半成品测试，再重新用 raw string 写测试。
+
+### 第二步：跑定向验证
+
+```bash
+luac -p internal/lua/adapters/trae.lua
+luac -p internal/lua/adapters/buddy.lua
+go test -tags luajit ./internal/provider/
+go test -tags luajit ./internal/lua/ -run 'Adapter|Buddy|Trae|Stream'
+```
+
+### 第三步：构建并部署临时验证版
+
+修复确认后：
+
+```bash
+go build ./...
+./deploy.sh
+```
+
+部署前确认：
+
+```bash
+sha256sum internal/lua/plugins/channels.lua
+sha256sum /etc/llmsproxy/plugins/channels.lua
+systemctl is-active llmsproxy
+```
+
+### 第四步：真实流式回归
+
+不要用 `stream:false` 测试流式问题。使用：
+
+```bash
+KEY='<从本机安全环境读取，不写入文档>'
+TOOL='{"type":"function","function":{"name":"get_weather","description":"Get weather","parameters":{"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}}}'
+
+curl --no-buffer --max-time 90 \
+  -H "Authorization: Bearer $KEY" \
+  -H 'Content-Type: application/json' \
+  -d "{\"model\":\"channels-codearts:openpangu-2.0-flash\",\"messages\":[{\"role\":\"user\",\"content\":\"Use the weather tool for Beijing.\"}],\"tools\":[$TOOL],\"tool_choice\":\"auto\",\"stream\":true,\"max_tokens\":64}" \
+  http://127.0.0.1:8081/v1/chat/completions
+```
+
+分别将 model 替换为：
+
+```text
+channels-codearts:openpangu-2.0-flash
+channels-trae:DeepSeek-V4-Flash
+channels-buddy:glm-5.3-flash
+```
+
+验收：
+
+- HTTP 200
+- `Content-Type: text/event-stream`
+- 不能出现 `tool_calls: {}`
+- 有 tool-call fragments 时最终必须是 `finish_reason: tool_calls`
+- CodeArts 当前应继续返回结构化工具调用
+
+### 第五步：CodeArts 刷新验收
+
+```bash
+curl -sS -X POST \
+  -H "Authorization: Bearer $KEY" \
+  -H 'Content-Type: application/json' \
+  -d '{"platform":"codearts"}' \
+  http://127.0.0.1:8081/api/plugins/http/channels/model/refresh
+```
+
+预期：
+
+```json
+{"ok":true,"results":[{"platform":"codearts","ok":true,"count":11}]}
+```
+
+再检查：
+
+```bash
+curl -sS -X POST \
+  -H "Authorization: Bearer $KEY" \
+  -H 'Content-Type: application/json' \
+  -d '{"platform":"codearts"}' \
+  http://127.0.0.1:8081/api/plugins/http/channels/model/list
+```
+
+预期：远端目录 11 个 + 兜底补位，已确认 404 的 4 个仍保持 `disabled=true`。
+
+## 8. 发布/安装包状态
+
+目前没有完成“全部安装包”发布闭环。
+
+已经发生的事情：
+
+- 核心包构建成功，但脚本实际使用了 `git describe`，产物版本类似：
+  `2.0.40~4-gc9f3453`
+- Linux GUI 包构建成功，版本为 `1.9.7`
+- Windows 主机 Wine 构建失败，生成了一个 264KB 的坏 `ModelRouter Setup 1.9.7.exe`，绝对不能发布
+- Docker Windows 构建也失败在拉取 Docker Hub 基础镜像 `golang:1.25` 网络超时
+- apt 无法找到 `wine32:i386`
+- 旧的 1.9.5/1.9.6 Windows 包是历史产物，不能作为当前版本
+
+不要把旧 Windows exe 当作新包发布。
+
+仓库已有 Docker 方案：
+
+```text
+cmd/gui/scripts/dist-win-docker.sh
+cmd/gui/docker/win-builder/Dockerfile
+```
+
+但当前 Docker Hub 网络不可用。若要继续做 Windows 包：
+
+1. 在可访问 Docker Hub 的构建机/CI 执行 Docker 方案；或
+2. 提供包含 Wine32/Wine64 的完整构建环境；或
+3. 暂时只发布 Linux/core 包，并明确说明 Windows 包未完成。
+
+版本脚本注意：
+
+- 核心包脚本使用位置参数：
+
+  ```bash
+  ./packaging/core-dist.sh 2.0.41
+  ```
+
+  不是 `VERSION=2.0.41 ./packaging/core-dist.sh`。
+- GUI 版本来自 `cmd/gui/package.json`，环境变量 `VERSION` 不会自动修改 Electron 产品版本。
+- 清理旧产物后再打包，并在全部包完成后统一计算 `SHA256SUMS`。
+
+## 9. 参考插件状态
+
+用户要求按：
+
+```text
+https://github.com/Ebony-Vinyl/dsh-our-free-model
+```
+
+逐厂商对照参考插件。
+
+当前本地路径：
+
+```text
+/tmp/pi-github-repos/Ebony-Vinyl/dsh-our-free-model
+```
+
+但该目录当前不是有效可读源码副本：
+
+- `vendor/channel-pack/src` 下面只有空目录；
+- `.git` 没有有效 `HEAD`/refs；
+- 无法从本地 Git 对象恢复源码；
+- 本地也没有 `buddy.ts`、`trae.ts`、`codearts.ts` 等文件。
+
+因此不能声称当前适配器是逐行照参考插件移植的。若要继续按用户要求完成，必须先重新 clone/恢复该 GitHub 仓库源码，确认以下文件实际存在后再对照：
+
+```text
+vendor/channel-pack/src/trae.ts
+vendor/channel-pack/src/buddy.ts
+vendor/channel-pack/src/codearts.ts
+vendor/channel-pack/src/*adapter*
+vendor/channel-pack/src/*model*
+```
+
+当前仓库代码只能作为临时协议修复基础，不能代替参考插件源码。
+
+## 10. 关键结论
+
+- Pi 是否发送流式请求，历史日志没有保存入站 `stream/tools` 字段，不能靠旧日志判断。
+- 本次真实 `stream:true + tools` 测试证明 ModelRouter 流式路径能工作：
+  - CodeArts：通过；
+  - Trae：通过但结束原因需归一；
+  - Buddy：通过但需过滤空 `{}`；
+  - MiniMax：凭据 401，独立问题。
+- 不要再用 `stream:false` 的手工请求推断 Pi 的问题。
+- CodeArts 刷新接口当前真实返回 `ok=true,count=11`，刷新链路不是当前阻塞点。
+- 当前最紧急的代码状态是修复 `provider_test.go` 的半编辑，然后验证未提交的统一流式修复。
+- 未经测试、未重新构建、未重新部署，不要继续改生产账号或刷新一次性凭据。

Makefile

@@ -2,7 +2,7 @@
 
 # ModelRouter - single-binary gateway + optional Electron desktop GUI (embedded core)
 export BINARY
-BINARY=llmsproxy
+BINARY=modelrouter
 export BUILD_DIR
 BUILD_DIR=build
 export GUI_DIR
@@ -20,7 +20,7 @@ all: build
 # ---- core binary (server users / embedded core) ----
 build:
 	@mkdir -p $(BUILD_DIR)
-	CGO_ENABLED=1 go build -tags luajit -trimpath -o $(BUILD_DIR)/$(BINARY) ./cmd/llmsproxy
+	CGO_ENABLED=1 go build -tags luajit -trimpath -o $(BUILD_DIR)/$(BINARY) ./cmd/modelrouter
 	@echo "Built: $(BUILD_DIR)/$(BINARY) ($(VERSION))"
 
 # ---- core distribution packages (deb + rpm + tar.gz) -----

README.md

@@ -4,103 +4,46 @@
 >
 > **AI-assisted**: 本项目使用 AI 辅助编程（代码与文档由 AI 协作完成，经人工复核）。
 
-面向多 llms 订阅者，部署在内网，实现一次配置多个服务共同使用的效果。支持指定模型或 auto 模式，按照配置的优先级选择可用模型提供服务。
+面向多 LLM 订阅者的内网网关：一次配置，多个服务共用一组上游额度。支持指定模型或
+`AUTO` 模式，按配置的优先级选择可用模型提供服务。
 
-> **轻量且增源无需重编译**：网关本体是单 Go 二进制（约 10MB，零运行时依赖）。新增/切换上游
-> 只需在 `config.yaml`（或 WebUI）加一个 `sources` 条目或挂一个 `.lua` 适配器——**不改 Go、
-> 不重编译**。源与适配器的改动经 WebUI 提交时即时生效（热更新）；直接编辑
-> `config.yaml` 或 `adapter_dir` 下的 `.lua` 文件则需要重启进程生效。
+网关本体是单个 Go 二进制（约 10 MB，零运行时依赖）。新增或切换上游只需在
+`config.yaml`（或 WebUI）加一个 `sources` 条目、或挂一个 `.lua` 适配器——**不改 Go、
+不重编译**。源与适配器的改动经 WebUI 提交时即时生效（热更新）；直接编辑
+`config.yaml` 或 `adapter_dir` 下的 `.lua` 文件需要重启进程生效。
 
-## 核心优势
+## 核心能力
 
-### 极致轻量
-
-- **单二进制**：编译后 8~12MB（`-s -w` strip 后约 8MB），零运行时依赖（仅依赖系统 libc），部署即用
-- **低内存占用**：与源数量相关，非与运行时长相关——单源 ~10MB、16 源生产实例 ~32–35MB（[实测分解与调优](#内存占用实测与调优)）
-- **日志按需加载**：统计数字来自**全量**审计日志（启动时流式扫一遍即释放，29MB/22 万行约 260ms），但原始记录不常驻内存——默认只加载首屏，下滚自动分页，CSV 导出流式写出（O(1) 内存），离页即释放
-- **零运行时依赖**：纯 Go + LuaJIT 静态链接，无需安装 Python/Node/Java 等运行时
-- **启动极快**：冷启动 < 200ms，热重载配置 < 10ms
-
-### 强大的多租户调度能力
-
-- **多密钥多租户**：支持无限密钥，每个密钥独立角色、模型范围、Token 配额、重置周期
-- **按模型配额**：每把 key 的每个模型单独配 token + 请求数配额与重置周期（小时/周/月/自定义 N 小时）；一个模型用满只影响该模型，同 key 其它模型照常；耗尽返 429 + `Retry-After` 并点名模型，admin key 永不受限
-- **AUTO 智能调度**：基于优先级档位的分级调度，同优先级源自动轮询负载均衡，故障自动毫秒级故障转移
-- **自愈冷却**：冷却上限 5 分钟，过半后放行 1 个探测请求，上游/额度恢复即刻回归轮询，无需等满冷却窗口
-- **Token 配额管理**：精确到模型级别的 Token 配额控制，支持小时/周/月/自定义小时周期自动重置
-- **同优先级源负载均衡**：同一优先级档位的多个源，请求自动 Round-Robin 均匀分发，故障毫秒级故障转移
-- **Source-Model 前缀路由**：支持 `source-model`/`source:model`/`source/model` 精确指定上游源
-
-### 生产级可靠性
-
-- **热加载配置**：WebUI 修改源/密钥/适配器/AUTO链即时生效，无需重启
-- **密钥加密落盘**：AES-256-GCM 加密存储 `api_key`、网关密钥、请求头，master.key 0600 权限保护
-- **实时源探测**：`GET {base}/models` 定期探测，毫秒级感知上游状态，不污染调度退避状态
-- **全链路审计**：HTTP 访问日志、登录/配置变更审计、请求记录 CSV 导出、密钥用量统计 CSV 导出
-
-### 灵活的协议适配
-
-- **LuaJIT VM**：每个适配器独立 VM + **弹性 worker 池**（按需扩缩容，空闲时自动回收），安全并发，Lua 脚本热加载无需重启
-- **Lua 适配器协议**：`transform_request` / `transform_response` / `transform_stream_chunk` 双向转换
-- **动态请求头钩子**：`build_headers(meta)` 支持 HMAC 签名、动态 Header 注入
-- **多模态透传**：`image_url` 等多模态内容在多源间无损透传，Anthropic/Gemini/Ollama 自动转换
-
-## 实现
-
-统一 OpenAI 兼容网关：把多个上游 LLM 源（DeepSeek、Qijiar、OpenAI、Anthropic、Gemini、
-Groq、Mistral、Ollama、KimiCode…）通过 **Lua 适配器** 做协议转换，对内网暴露
-一个标准 `OpenAI Chat Completions` 接口（`/v1/chat/completions` + `/v1/models`），
-支持单次结算、SSE 流式、**AUTO 模型路由**与**多模态**透传。
-
-从 [HomeAgent](https://gitcode.com/JianFeeeee/HomeAgent) 的多源 LLM 适配层
-（`internal/agent/api/provider.go` + `internal/lua/adapters/*`）抽离而来并独立演进。
-
-## 特性
-
-- **多源**：一个进程内配置任意多个上游源，按请求的 `model` 自动路由。
-- **AUTO 模式**：`default_model: AUTO` 时按优先级页保存的 AUTO 链档位逐档调度并发请求。
-- **统一输出**：所有源对外都是 OpenAI 格式（含 `reasoning_content`、`tool_calls`、`usage`）。
-- **生图**：`POST /v1/images/generations`，`kind: image` 的模型独立路由。
-- **多模态**：`content` 数组（`image_url` 等）在多源间无损透传，Anthropic/Gemini/Ollama 自动转换。
-- **LuaJIT VM**：基于 golua 绑定的 LuaJIT，每个适配器独立 VM + worker 池，安全并发。
-  池是 **弹性** 的：`Σ max_concurrent` 只是上限而非预分配，状态按需创建、空闲时自动回收，
-  因此空载网关几乎不持有 Lua 状态（启动 0 个）。
-  - 扩容步长由适配器最大并发决定：`clamp(ceil(max/8), 1, 8)`，且仅在**真实并发争用**
-    （所有现有状态均已占用）时批量预热，顺序流量始终只用 1 个状态。
-  - 缩容步长由当前连接数决定：`clamp(ceil(冗余/(1+占用)), 1, 冗余)`——无连接时一轮收到
-    常驻下限，忙时每轮只释放 1 个，保护热路径。WebUI「适配器」页展示实时池指标。
-- **thinking 预算**：请求里的推理开关/预算归一为中立的 `thinking:{enabled,budget_tokens}`，
-  适配器映射到各自协议（Anthropic `thinking.budget_tokens`、Gemini `thinkingConfig`、
-  DeepSeek `thinking.type="disabled"`）。
-- **Lua 适配协议**：每个源挂一个 `.lua` 适配器，完成 `transform_request` /
-  `transform_response` / `transform_stream_chunk` 双向转换，协议差异全在 Lua 层。
-- **签名 / 请求头钩子**：适配器可定义 `build_headers(meta)`，在 Go 发 HTTP 前
-  动态注入/签名请求头——用于云端 API 校验调用方 app（如 KimiCode 只放行特定
-  agent）。提供 `hmac_sha256_hex` / `sha256_hex` / `base64_encode` 等签名辅助。
-- **WebUI**：内置管理界面，可在线查看/新增/编辑上游源与模型、配置 AUTO 优先级链、
-  管理密钥模型范围，写入运行时文件持久化。
-- **密钥加密存储**：运行时文件中的上游 `api_key`、自定义请求头值、网关 key 均以
-  AES-256-GCM 加密落盘（`master.key` 独立 0600，或 `LLMS_PROXY_MASTER_KEY`）。
-- **实时源探测**：状态页探测各源可达性（`GET {base}/models`，失败回退最小请求），
-  不污染正常调度的退避状态，错误信息在 UI 可悬停查看。
-- **鉴权**：网关自身用 `gateway_keys` 校验客户端 Bearer key；与上游各自的 key 相互独立。
-- **流式**：SSE `chat.completion.chunk`，含角色首包与 `[DONE]` 收尾。
-- **全链路审计**：HTTP 访问日志、登录/配置变更审计、请求记录 CSV 导出、密钥用量统计 CSV 导出
-- **AES-256-GCM 加密存储**：运行时文件敏感字段加密落盘，master.key 0600 权限，支持环境变量注入主密钥
+- **统一接口**：对内暴露 OpenAI 兼容的 `/v1/chat/completions`、`/v1/models`、
+  `/v1/images/generations`；所有源对外都是同一套格式（含 `reasoning_content`、
+  `tool_calls`、`usage`）。
+- **多源 + AUTO 链**：一个进程内任意多个上游源；`default_model: AUTO` 时按优先级页
+  保存的档位逐档调度，源故障自动降级并冷却，恢复后半冷却探测自愈。
+- **协议适配在 Lua 层**：每个源挂一个 `.lua` 适配器，完成请求/响应/流式分片的双向
+  转换，并可用 `build_headers(meta)` 注入签名头。协议差异不进 Go 内核。
+- **插件独立**：业务扩展（渠道接入、计费、额度巡检）是外置 Lua 插件，由插件目录加载，
+  不编进二进制、不随发行包分发；插件只通过 `modelrouter` SDK 与内核交互。
+- **多租户密钥**：网关 key 与上游 key 相互独立，按 key 限定模型范围、AUTO 链与配额。
+- **加密存储**：上游 key、自定义请求头、网关 key 均以 AES-256-GCM 落盘，
+  `master.key` 单独 0600 权限（或由 `MODELROUTER_MASTER_KEY` 注入）。
+- **可观测**：状态页实时探测源可达性；HTTP 访问日志、配置变更审计、请求与用量 CSV 导出；
+  写操作不落在请求热路径上。
 
 ## 快速开始
 
 ```bash
-GOMODCACHE=... GOPROXY=off go build -tags luajit -o llmsproxy ./cmd/llmsproxy
-./llmsproxy -config config.yaml    # 首次运行自动生成默认配置并打印随机 admin key
+# 需要 LuaJIT 开发库；-tags luajit 是必须的
+GOMODCACHE=... GOPROXY=off go build -tags luajit -o modelrouter ./cmd/modelrouter
+
+# 首次运行会生成默认配置，并在日志里打印随机 admin key
+./modelrouter -config config.yaml
 ```
 
-> 依赖 [golua](https://github.com/aarzilli/golua)（LuaJIT 绑定）。**必须**带 `-tags luajit`
-> 构建，否则默认走内置超集 gopher-lua 路径（行为略有差异）。
+> 依赖 [golua](https://github.com/aarzilli/golua)（LuaJIT 绑定）。不带 `-tags luajit`
+> 会走 gopher-lua 路径，行为略有差异。
 
-> **配置不入库**：仓库不携带任何 config 文件（配置文件含密钥）。二进制首次运行时
-> 会在 `-config` 指定的位置生成默认配置：随机 admin key（打印在启动日志中）、
-> 仅绑定 `127.0.0.1:8080`。首次登录后请在 WebUI「密钥」页更换管理员密钥。
+> **配置不入库**：仓库不携带任何 config 文件（其中含密钥）。默认配置只绑定
+> `127.0.0.1:8080`，首次登录后请在 WebUI「密钥」页更换管理员密钥。
 
 ```bash
 # 无 key -> 401
@@ -117,526 +60,73 @@ curl -N -H "Authorization: Bearer $KEY" \
      http://127.0.0.1:8080/v1/chat/completions
 ```
 
-任何 OpenAI SDK 把 `base_url` 指到网关地址、`api_key` 用任意一个已创建并授权的网关 key 即可。
+任何 OpenAI SDK 把 `base_url` 指向网关、`api_key` 换成任一已授权的网关 key 即可。
 
-## 配置
+## 文档
 
-首次运行在 `-config` 路径自动生成默认配置（默认 `./config.yaml`），核心字段：
+| 主题 | 文档 |
+|---|---|
+| 实现与完整特性清单 | [docs/features.md](docs/features.md) |
+| 配置：`config.yaml`、自动探测模型、多密钥 | [docs/configuration.md](docs/configuration.md) |
+| 路由：模型路由、AUTO 链、冷却与半冷却、源与适配器、源模板 | [docs/routing.md](docs/routing.md) |
+| 适配器：Lua 协议、内置适配器清单 | [docs/adapters.md](docs/adapters.md) · [docs/lua-adapters.md](docs/lua-adapters.md)（[EN](docs/lua-adapters-en.md)） |
+| 插件：SDK 契约、钩子、UI 扩展、HTTP 端点 | [docs/plugins.md](docs/plugins.md) |
+| 桌面端：Headless / Desktop 两种产物、构建与运行 | [docs/gui.md](docs/gui.md) |
+| 性能：实测内存占用与调优、thinking 预算 | [docs/performance.md](docs/performance.md) |
+| WebUI 使用 | [docs/webui.md](docs/webui.md) |
+| 开发：目录职责、测试、分支与发布纪律 | [docs/development.md](docs/development.md) · [docs/git-workflow.md](docs/git-workflow.md)（[EN](docs/git-workflow-en.md)） |
+| 部署与迁移 | [docs/deployment.md](docs/deployment.md) |
 
-```yaml
-listen: 127.0.0.1:8080        # 网关监听地址（建议绑内网/回环）
-gateway_keys: [sk-gw-<随机>]  # 初始 admin 密钥（每次安装随机生成，仅首启写入运行时存储用）
-default_model: AUTO           # model 无法路由时自动按优先级链选源
-adapter_dir: adapters         # Lua 适配目录；目录不存在时首启创建并 seed 内置，存在则只读
-runtime_file: runtime.json    # WebUI 编辑的源/密钥/AUTO 链持久化到此文件
+## 从 llmsproxy 改名
 
-sources:
-  - name: deepseek
-    base_url: https://api.deepseek.com
-    api_key: sk-...            # 也可用 api_key_env: SOME_ENV 引用环境变量（不落盘明文）
-    adapter: deepseek
-    max_concurrent: 8
-    models:
-      - id: deepseek-v4-flash
-        priority: 100          # YAML 源首次启动会 seed 进 AUTO 链的初值
-        kind: chat
-      - id: deepseek-v4-pro
-        priority: 60
-        kind: chat
-    # 静态请求头（比适配器默认优先）
-    headers: { X-Tenant: prod }
-    # 传给 Lua build_headers 的透传元数据
-    meta: { app_id: x, app_secret: y }
-    temperature: 0.7
-    max_tokens: 4096
-    timeout: 120s              # 请求超时，默认 120s
-```
+产品名早已是 ModelRouter，二进制、配置目录、服务名与环境变量现已统一为
+`modelrouter`：
 
+| 旧 | 新 |
+|---|---|
+| `llmsproxy`（二进制） | `modelrouter` |
+| `/etc/llmsproxy/` | `/etc/modelrouter/` |
+| `llmsproxy.service` | `modelrouter.service` |
+| `LLMSPROXY_*` | `MODELROUTER_*` |
 
-### 自动探测模型列表
-
-不想手抄模型 id 时，只填 URL 和 key，让网关去扫：
+已部署的机器用迁移脚本升级（**默认 dry-run**，复制不移动，可回滚）：
 
 ```bash
-# 只探测，不写任何东西
-curl -X POST http://127.0.0.1:8080/api/sources/discover \
-  -H "Authorization: Bearer sk-gw-…" -H "Content-Type: application/json" \
-  -d '{"base_url":"https://api.vendor.com/v1","api_key":"sk-…"}'
-# → {"ok":true,"base_url":"…","endpoint":"","models":["gpt-x",…],
-#    "image_models":["gpt-image-2"],"models_path":"/models","latency_ms":1387,…}
-
-# 探测后直接落库
-curl -X POST http://127.0.0.1:8080/api/sources/discover/apply \
-  -H "Authorization: Bearer sk-gw-…" -H "Content-Type: application/json" \
-  -d '{"base_url":"https://api.vendor.com/v1","api_key":"sk-…","name":"vendor"}'
+sudo ./packaging/migrate-to-modelrouter.sh              # 预演
+sudo ./packaging/migrate-to-modelrouter.sh --apply      # 执行
+sudo ./packaging/migrate-to-modelrouter.sh --rollback --apply
 ```
 
-WebUI 源编辑框里也有「自动探测模型」按钮：填好 URL 和 key 点一下，列表自动
-填好，仍需点保存才生效。
-
-几点值得知道：
-
-- **URL 形态不用猜。** 厂商对「`/v1` 放 base 还是放 path」没有共识，探测会
-  依次尝试若干候选并返回答对的那个，同时给出归一化后的 `base_url` +
-  `endpoint`（`endpoint` 平时留空，因为 openai adapter 自带 `/chat/completions`）。
-- **`image_endpoint` 是另一套规则。** 网关的图像默认路径是**绝对**的
-  `/v1/images/generations`，而 chat 默认是相对的。所以 `base_url` 带 `/v1` 时
-  必须显式给 `image_endpoint: /images/generations`，否则图像请求 404。
-  探测会替你填好。
-- **图像模型只看上游自己声明的 modality**，绝不从模型名猜。`gpt-image-2`
-  这类 id 也出现在同时提供对话服务的目录上，猜错会让 AUTO 链选中一个只支持
-  图像生成的模型，其后每一层都被它拖垮。
-- **不会逐模型探活。** 34 个模型的目录跑一轮要花钱且耗时数分钟，`/models`
-  是足够便宜的信号；模型是否真能应答留给正常请求去发现。
-- **对已存在的源，必须显式 `replace_models: true`** 才会覆盖模型列表。重新
-  探测顺手抹掉为 AUTO 链手工调过的 priority，是最难排查的一类数据丢失。
-- **`api_key` 留空表示沿用已存值**，不会把凭据清空。
-
-
-`config.yaml`、`runtime.json`、`master.key`、`adapters/` 均不提交版本库（见
-`.gitignore`），请随备份带走——`master.key` 丢失后运行时密文无法解密。
-
-运行文件生成在 `-config` 所在目录：`adapter_dir` / `runtime_file` 自动指向
-配置文件旁边的 `adapters/` 与 `runtime.json`，从任意工作目录启动都可用。
-
-运行时文件（`runtime_file`）中的敏感字段自动加密：
-
-- 加密算法 AES-256-GCM，格式 `enc:v1:<base64>`。
-- 主密钥来源：环境变量 `LLMS_PROXY_MASTER_KEY`（64 位 hex）；否则读取
-  `runtime_file` 同目录的 `master.key`；都不存在时首次启动自动生成 `master.key`（0600）。
-- 升级时旧明文文件自动兼容：首次运行正常读取，任何 UI 保存操作触发全文件加密迁移。
-- 注意：**master.key 丢失后密文无法解密**，请随配置一起备份；切勿提交到版本库。
-
-### 网关密钥（多密钥）
-
-网关鉴权采用「多密钥 + 角色 + 模型范围」架构，密钥持久化在 `runtime_file` 的
-`keys` 字段（加密存储）：
-
-- `gateway_keys` 配置只是**初始 admin 密钥种子**：首次启动迁移为运行时 admin
-  key，之后不再参与鉴权管理。
-- **重要：首次启动后请在 WebUI「密钥」页更换管理员密钥**——初始密钥明文写在
-  `config.yaml` 里（每次安装随机生成），继续使用存在被盗风险；用新密钥登录后删除初始密钥。
-- WebUI **密钥页**可创建/删除密钥；每个密钥可指定 `admin`（管理全部）或
-  `user`（仅看自己的 key）角色，并配置**模型范围**（模型 + 源 + token 配额 +
-  重置周期）。
-- 客户端用任意一个已授权的密钥明文作为 Bearer（`Authorization: Bearer <key>`）。
-- 删除密钥即从运行时存储移除，立即失效。
-
-#### 密钥用量配额
-
-**配额按模型单独设置**：每个密钥的 `models[]` 里，每一条模型范围各自带一份
-token 配额与请求数配额。一个模型用满只影响该模型，同一密钥的其它模型照常工作。
-
-```yaml
-keys:
-  - key: sk-gw-<hex>
-    role: user
-    name: agent-alice
-    models:
-      - model: deepseek-v4-flash
-        token_quota: 1000000   # 本周期内该 key 用这个模型的 token 预算
-        req_quota: 20000       # 本周期内的请求次数
-        period: nhour          # "" | hour | week | month | nhour
-        hours: 6               # 仅 nhour
-      - model: AUTO
-        token_quota: 5000000   # AUTO 也是一条独立配额
-        period: day            # ← 这种写法会被拒绝（词表只有 hour/week/month/nhour）
-      - model: kimi-k3        # 未列配额 = 无限
-```
-
-- **没有「整钥总配额」**：这是刻意的设计。整钥总额会让「把 A 模型的额度挪给
-  B 模型」变成一次全局重分配；按模型独立计费则每个模型各自可控，运维可以
-  看出哪个模型吃掉了预算。
-- `period` 词表：空 = 永不过期（累计总量），`hour` / `week` / `month` = 固定窗口，
-  `nhour` + `hours` = 自定义小时数。**拼错的周期在写入时就被拒**，不会静默变成
-  永不过期。
-- 配额严格按密钥隔离，且**同一密钥内按模型隔离**：A 密钥用满 `m1` 不会消耗
-  B 密钥的额度，同一密钥的 `m2` 也不受影响。
-- 配额统计含聊天、流式、生图，跨重启从审计日志回放（保留 40 天，覆盖最长的
-  month 窗口）。
-- 配额耗尽返回 **429 + `Retry-After`**（`rate_limit_exceeded`），消息里点名是哪个
-  模型用满了，客户端可等窗口重置后自动恢复；模型越权才是 403。
-  **admin 密钥永不受配额限制**（其 scope 上的配额也不强制），避免把管理员
-  锁在门外。
-- 窗口用量按整点小时分桶统计，实际释放比配置窗口最多晚 1 小时（配额宁可晚释放
-  也不超发）。
-- `PUT /api/keys/{key}` 提交 `models` 即同时提交它们的配额（配额就是 scope 的一部分，
-  不存在会与模型列表脱节的第二份预算）。显式 `0` = 解除该模型的限制。
-
-##### 配额拒绝 vs 容量拒绝：两种「拒绝」含义不同
-
-配额在**入口**检查，不占用任何上游槽位，因此是廉价拒绝；容量不足则要走
-`busyWait` 有界等待后才返回 503。实测（容量 8、0.6s/请求、100 并发）：
-
-| 情况 | 状态码 | 延迟 | 是否打到上游 |
-|---|---|---|---|
-| 配额用尽（`req_quota` 50 → 100 并发） | `429 rate_limit_exceeded` + `Retry-After` | **19–21ms** | 否 |
-| 容量打满（同一拓扑，未限配额） | `503 upstream_error` | ~2.6s（`busyWait`） | 超出部分否 |
-
-客户端因此可以区分「等窗口重置」与「等上游腾出容量」：前者只需等待，后者
-通常该降并发或换源。
-
-##### 配额的开销
-
-配额桶按 (密钥, 模型, 整点小时) 分桶保留 40 天，实测（AMD 7840HS）：
-
-- 窗口查询按窗口长度而非保留总量扫描：1h 窗口 49ns、24h 窗口 55ns、
-  30d 窗口 3.9μs（此前全扫保留总量，960 桶时 5.9μs）。
-- 记录一条请求：283ns、3 分配（与引入配额前相同，分配来自 ring buffer）。
-- 内存：生产形态（7 密钥 × 8 模型 × 2 源 × 满 40 天 retention）约 **3.7MB**。
-  按源 pin 的 `source::model` 桶**惰性创建**——只有当某条配额真的 pin 了
-  某个源时才维护，否则每条记录多写一份桶，在 20 密钥 × 8 模型 × 3 源下会
-  多占 18MB。
-  代价：配置 pinned 配额之前的历史用量无法事后按源拆分，所以 **pinned 配额的
-  首个窗口可能少算**。
-
-```bash
-# 配额耗尽时客户端看到（点名了具体模型）
-HTTP/1.1 429 Too Many Requests
-Retry-After: 2100
-{"error":{"type":"rate_limit_exceeded","message":"token quota exceeded for \"deepseek-v4-flash\" (5000000/5000000)"}}
-```
-
-### 模型路由
-
-`/v1/chat/completions` 的 `model` 解析顺序：
-
-1. `source/model` 或 `source:model` 前缀 → 指定源；
-2. 精确匹配某个源的 `model`；
-3. 配置 `default_model: AUTO` 时 → 走优先级页保存的 AUTO 链（见下）；
-4. 否则回落（返回错误）。
-
-任何 OpenAI 客户端，只要 `model` 设为某个源的 `name/任意名`，即可锁定走该源；
-设为 `AUTO`（或网关配了 `default_model: AUTO`）即自动按优先级链选源。
-
-### AUTO 链（优先级页）
-
-AUTO 调度**只**由优先级页保存的规则决定，源配置里的 `models[].priority` 数字不再
-参与调度、也不再显示。**聊天与生图是两套独立的链**，在优先级页面通过「聊天 / 生图」
-按钮切换编辑：
-
-- **聊天链**（`runtime_file` 的 `auto` 字段）：服务 `POST /v1/chat/completions`
-  的 `model: AUTO` 请求；只接受 `kind: chat` 模型，配置在链里的生图槽位会被忽略。
-- **生图链**（`runtime_file` 的 `auto_image` 字段）：服务 `POST /v1/images/generations`
-  的 `model: AUTO` 请求；只接受 `kind: image` 模型。未配置生图链时，AUTO 生图回落
-  到「所有含生图模型的源、按注册顺序」的旧行为，兼容旧配置。
-
-每条规则 = 一个「槽位」：`{ model, source, tier, token_quota, period, hours }`。
-
-- `tier` 表示优先级档位：同一档的模型并排、共享该优先级；档位从上到下递减。
-- 同一模型可配置多个槽位（如 A 源低配 → B 源低配 → A 源高配 → B 源高配），按档位顺延。
-- `token_quota` > 0 时该槽位在重置周期内用满即顺延到下一槽位；`period` 支持
-  `hour` / `week` / `month` / `nhour`（配合 `hours`），空 = 不限。
-- 同 tier 内按「偏好分」排序，失败槽位冷却后自动跳过；冷却 / 配额耗尽 / hard
-  错误会顺延到下一 tier，全部失败时返回 503 并附上每档失败摘要。
-
-### 冷却与半冷却探测（自愈调度）
-
-槽位失败后进入指数退避冷却（5s → 10s → 20s …，上限 **5 分钟**）。冷却不是全时段
-硬闸：
-
-- **前半段完全静默**：刚失败的槽位不接任何流量，避免对故障上游持续施压。
-- **后半段放行 1 个探测**：窗口过半后，同一 `(源, 模型)` 最多允许 **一个** 请求作为
-  探测通过。探测槽数 = `clamp(max_concurrent / 10, 1, 2)`——`max_concurrent: 10`
-  的源恰好放行 1 个探测请求。
-- **探测排在最后**：同 tier 内探测候选永远排在健康候选之后，只有没有正常槽可用时
-  才承接真实流量，因此探测不会抢走可用容量。
-- **探测即真实请求**：成功则立即清零失败计数与冷却，槽位当场回到正常轮询——
-  上游恢复/额度恢复后不必等满整个冷却窗口。失败则重开一个新窗口，下次探测顺延到
-  新窗口的中点，不会连续重试。
-
-分类冷却策略：
-
-| 情况 | 冷却 | 说明 |
-|---|---|---|
-| 传输错误 / 5xx | 5s→10s→20s…，上限 5min | 指数退避 |
-| 401 / 403 凭据错误 | 10min | 换 key 后由下一次探测自动接回 |
-| 429 限流 | 30s 固定 | 「太快」不等于「坏了」，不进指数阶梯 |
-| 配额耗尽（`insufficient_quota` / `余额不足` 等） | 对齐配额窗口，上限 30min | **不进指数阶梯、不涨失败计数**，额度恢复即可用 |
-
-WebUI「优先级」页的槽位健康标签会区分 `冷却` / `待探测` / `探测中`，悬停可看窗口
-起点、探测放行时刻与完全恢复时刻。
-
-### 源（Source）与适配器（Adapter）的关系
-
-- **源** 是到某个上游的连接描述：`name`、`base_url`、`api_key`、模型列表与优先级。
-- **适配器** 是协议转换逻辑（Lua 脚本）：把统一 OpenAI 格式请求转成上游原生格式，
-  再把上游响应/流式分块转回统一格式。
-- 一个适配器可被多个源复用（如 `openai.lua` 同时服务多个 OpenAI 兼容站点）；
-  同一个源也可以切换不同适配器（改 `adapter` 字段即可）。
-- 源决定“连谁、暴露哪些模型”，适配器决定“怎么对话”——二者在 `sources[]` 条目中
-  通过 `adapter` 字段关联。
-
-加载流程（`internal/core` 的 `Core` 负责装配）：
-
-1. 启动时 `lua.NewVM(adapter_dir)` 加载全部适配器：内置适配器（编译期 embed）+
-   `adapter_dir` 下的同名覆盖文件；内置文件写在代码内，覆盖文件要求更高优先级。
-2. `config.Load` 读取 `config.yaml`，`config.NewStore(runtime_file)` 读 WebUI 改动的
-   运行时源，二者按名称合并成完整源列表。
-3. `rebuildRegistry` 为每个源创建 `provider.Provider`（持有目标适配器），并按源的
-   并发上限配置适配器 worker 池**上限**（不预分配，见下）。
-4. 请求进来时 `Registry` 按 `model` 路由到 Provider，Provider 调适配器
-   `transform_request` → HTTP 发送 → `transform_response` / `transform_stream_chunk`。
-
-WebUI 上的"新增/编辑源"、"上传 Lua 适配器"、"改 AUTO 优先级链"与"密钥模型范围"
-都即时生效（写入运行时文件或 `adapter_dir` 后重新装配，无需重启）；直接编辑
-`config.yaml` / `adapter_dir` 下的文件则需要重启进程才会重新加载。
-
-在 WebUI 删除源/适配器时：运行时（WebUI 创建）的源与 `.lua` 文件会真正移除；
-`config.yaml` 中定义的基源无法改写配置文件，采用删除标记隐藏（重启后仍隐藏），
-在 UI 里重新添加同名源即可恢复。
-
-### 源模板（多 key 负载均衡）
-
-多个上游 key 共享同一套配置（URL / 适配器 / 模型列表 / 并发 / rpm）时，逐一复制
-整段 source 太冗余。**源模板**保存除 `name` 和 `api_key` 以外的全部源字段，
-从模板一键创建多个仅 key 不同的源，天然支持多 key 负载均衡（每个展开源独立调度、
-独立冷却、独立健康状态）。
-
-- WebUI「源」页右上角「**模板管理**」按钮：列出全部模板，可编辑 / 删除 / 新建。
-- 添加源弹窗右上角两个按钮：
-  - **从模板创建** → 选模板 → 表单自动填充（name + key 仍需你填）；
-  - **存为模板** → 把当前表单非 key/name 字段存为新模板。
-- 模板持久化在 `runtime_file` 的 `source_templates` 字段，与运行时源一样热管理，
-  改完即时生效。模板本身不含密钥，可以安全共享 / 纳管。
-
-模板不会自己变成源——它只是「配方」。真正承担流量的是从模板创建出来的、带 key
-的具体源。
-
-### 内存占用（实测与调优）
-
-内存不是一个定数：它随**配置的源数量**增长（每个源一个 `http.Transport` 与连接池、
-一组健康状态），与运行时长无关。本机实测（Linux x86_64，12 核）：
-
-| 部署形态 | 启动 RSS | 稳态 RSS |
-|---|---|---|
-| 1 源 / 1 适配器（最小配置） | ~4 MB | ~10 MB |
-| 1 源 + 29 MB 历史审计日志 | ~19 MB | ~20 MB |
-| **16 源 / 13 适配器 / 59 模型（本机生产）** | ~28 MB | **~32–35 MB** |
-
-> 历史参考：本项优化前同一生产配置为 **~105 MB**。降幅来自三处：审计日志的**原始记录**
-> 不再常驻内存（约 25 MB；统计聚合仍扫全量，但扫完即释放）、Lua 状态池不再单调增长、
-> 以及下面两个运行时开关。
-
-内存构成（生产实例分段测量，`/proc/<pid>/smaps`）：
-
-| 区域 | RSS | 说明 |
-|---|---|---|
-| Go 堆 | ~14 MB | provider/registry/scheduler 结构 + 连接池缓冲 |
-| 其他匹名（线程栈 / LuaJIT chunk / runtime） | ~12 MB | 与线程数、已加载适配器数相关 |
-| 二进制 text+rodata | ~8 MB | 映射的可执行文件页（只读、可被内核回收） |
-| Go runtime 预留 | ~4 MB | VSZ 上看到的 2 GB+ 是地址空间预留，不占物理内存 |
-| 共享库 | ~3 MB | libc / libluajit / libm |
-
-**两个推荐的部署开关**（只适用于环境变量，无需改代码）：
-
-```ini
-# /etc/systemd/system/llmsproxy.service
-Environment=GOGC=50
-Environment=MALLOC_ARENA_MAX=2
-```
-
-- `MALLOC_ARENA_MAX=2`：LuaJIT 的分配走 cgo → glibc malloc，glibc 默认允许 `8×nproc`
-  个 per-thread arena，每个碰到 malloc 的 OS 线程会占用一个（各约 1 MB，且**不归还给系统**）。
-  实测从 8–15 个 arena（约 7–12 MB）降到 **0**。
-- `GOGC=50`：把 Go 堆增长目标减半。**单独使用无效**（省下的堆会立即被更多 glibc
-  arena 吃掉，实测 20.3 → 21.5 MB 反而变大），必须与 `MALLOC_ARENA_MAX` 配合，
-  两者同时开启才降 **~19%**。网关是 I/O 密集型（本机 9 小时仅消耗 1min10s CPU），
-  多出的 GC 周期与它的空闲 CPU 相比可忽略。
-
-> 桌面版（Electron）已在 `cmd/gui/main.js` 里默认为内嵌核心注入这两个开关；手动 export 同名
-> 环境变量可覆盖。如果你用自己的 systemd unit / 容器，建议照上面加上。
-
-可选：`GOMEMLIMIT=48MiB` 作软上限，实测再省 ~1 MB；代价是逐近上限时 GC 转激进，
-温和场景不必开。
-
-### thinking 预算
-
-请求体带 `"thinking": {"enabled": false}` 时关闭推理；`"thinking": {"enabled": true,
-"budget_tokens": 4096}` 指定预算。内核只定义这个**中立**语义字段，映射到具体协议由
-适配器负责（Anthropic `thinking.budget_tokens`、Gemini `thinkingConfig.thinkingBudget`、
-DeepSeek `thinking.type = "disabled"`）。内核不再向适配器泄漏 `disable_thinking` /
-`extra_body` 这类私有字段。
-
-### WebUI
-
-内置管理界面（`GET /`），登录后可在浏览器完成：
-
-- **状态页**：源在线状态（实时探测 + 悬停看错误）、模型/源/key 用量统计、请求记录，
-  支持按时间范围导出 CSV；点击模型可生成 pin 到该模型的连接配置。
-- **对话页**：流式/非流式调试。
-- **密钥页**：创建/编辑网关 key，为每个 key 配模型范围（模型 + 源 + token 配额 + 周期），
-  管理员管理全部 key，用户只看到自己的 key。每个模型砖块显示自己的配额徽标
-  （如 `1.0K 77×·1h`，未设配额显示 `∞`），点开可编辑该模型的 token 配额、
-  请求数配额与重置周期 —— 配额按模型独立生效，一个用满不影响同一 key 的其它模型。
-  「我的密钥」页对用户展示本 key 的模型范围与各自配额。
-- **优先级页**：拖拽积木配置 AUTO 链档位。
-- **源页**：在线增删改上游源（API key 等敏感字段加密落盘）。
-- **适配器页**：上传 / 删除 Lua 适配器脚本。
-
-改动写入 `runtime_file`（重启仍生效）。
-
-## Lua 适配器协议
-
-完整 API 见 **[Lua 适配器 API 文档](docs/lua-adapters.md)**（[English](docs/lua-adapters-en.md)）。
-
-每个适配器是一个返回 table 的 Lua 脚本（`internal/lua/adapters/<name>.lua`），
-加载时可被 `adapter_dir` 下的同名脚本覆盖——**改 Lua 脚本无需重编译**，
-重启即生效；通过 WebUI 上传的适配器与在线编辑的源配置则即时生效。
-
-```lua
-return {
-  name = "mysrc",
-  version = "1.0.0",
-  endpoint = "/chat/completions",        -- 上游路径（可被 source.endpoint 覆盖）
-  headers = { ["X-Static"] = "v" },      -- 静态默认请求头（build_headers 缺省时使用）
-
-  -- 请求转换：把 OpenAI 格式 req 转成上游原生格式，返回字符串
-  transform_request = function(raw_json) ... end,
-
-  -- 响应转换：把上游原生响应转成统一格式字符串
-  --   { content, reasoning_content, finish_reason, token_usage{...}, tool_calls[{...}] }
-  transform_response = function(raw_json) ... end,
-
-  -- 流式分块转换：把上游 SSE data 转成 { content, done, ... }，返回 "" 则跳过
-  transform_stream_chunk = function(raw_chunk) ... end,
-
-  -- [可选] 动态请求头/签名钩子
-  -- meta = { url, method, body, api_key, timestamp, source={ name, meta={...} } }
-  build_headers = function(meta) return { ["X-App-Sign"] = sign } end,
-
-  -- [可选] 出向序列化（契约 v2）：上游不是 OpenAI 协议时接管写给客户端的字节。
-  -- 不声明则沿用内核的 OpenAI 兜底；声明即完全取代，内核不再混合字段。
-  serialize_response = function(result) return body_str end,
-  serialize_open_preamble = function(result) return bytes end,
-  serialize_chunk = function(delta) return bytes end,
-  serialize_close = function(result) return bytes end,
-}
-```
-
-内置信号辅助：`hmac_sha256_hex(key, data)`、`sha256_hex(data)`、`base64_encode(s)`、
-`tohex(s)`、`json.encode/decode`、`log(level, msg)`。
-
-### 内置适配器
-
-`openai` `deepseek` `anthropic` `gemini` `github` `groq` `mistral` `ollama` `kimicode`
-`qwen38` `sensenova` `trae`。
-
-`anthropic`/`gemini`/`ollama` 适配器内置多模态转换（`image_url` → 各自上游格式）；
-推理开关由中立的 `thinking` 字段表达，`deepseek` 适配器把它映射为
-`thinking.type = "disabled"`。
-
-**kimicode** 是展示 `build_headers` 的样例：云端校验调用方 app，需要按
-`meta.app_secret` 对时间戳+URL+请求体哈希做 HMAC 签名并附 `X-App-Sign` 等头。
-配好 `sources[].meta.{app_id, app_secret, app_agent}` 即可。
-
-## 桌面 GUI（Electron）
-
-可选独立产物，面向内网桌面端用户：**内嵌 ModelRouter 核心**（Clash Verge 形态），
-一键启停、系统托盘常驻、开机自启、静默启动，内容区一比一内嵌完整 WebUI
-（状态/对话/密钥/优先级/源/适配器 六个页面，免登录）。服务器用户继续使用纯 Go 二进制。
-
-### 两种安装包：Headless 与 Desktop
-
-发布到 Release 的安装包分为两类，面向不同用户：
-
-- **Headless（服务器版）**：纯核心二进制（无 Electron），配置驱动，适合服务器、
-  容器、systemd 等无人值守场景。发布包含 `deb` + `rpm`（含 systemd unit、内存调优
-  与示例配置）与 `tar.gz`（二进制 + `config.example.yaml` + `adapters/` + README），
-  Windows 为 `.zip`。
-- **Desktop（桌面版）**：Electron GUI 安装包，内嵌核心，适合个人桌面日常使用。
-
-### 选择 GUI 还是纯后端
-
-| 场景 | 推荐 | 理由 |
-| ------ | ------ | ------ |
-| 服务器 / 内网网关 / 无人值守常驻 | **Headless**（单二进制） | 极轻量（二进制 ~8MB；RSS 随源数量，单源 ~10MB、16 源 ~32–35MB，见[内存占用](#内存占用实测与调优)），零依赖单进程，直接跑在 systemd/任意容器里，远程管理 |
-| 个人桌面日常使用 / 多设备内网共享 | **Desktop**（GUI） | 免登录内嵌 WebUI、系统托盘一键启停、开机自启、静默后台，适合不懂命令行的使用者 |
-| Windows 桌面 | **Desktop** | 纯后端在 Windows 上需自行注册服务，GUI 提供原生托盘/自启体验 |
-| CI 一键出三平台安装包 | **Desktop 打包脚本** | `make gui-*` 系列出 deb/AppImage/NSIS，可进发行流水线 |
-
-两者完全同源：GUI 内嵌的就是纯后端的同一份 `llmsproxy` 核心（LuaJIT 版），
-配置/适配器格式完全一致，可随时互换。
-
-### 构建与运行
-
-```bash
-cd cmd/gui
-npm install
-make gui          # 或 npm run dev —— 开发运行（需先 make build 生成 bin/llmsproxy）
-make gui-deb      # 本机构建 deb（分享给其它 Linux 用户）
-make gui-dist     # 构建 deb + rpm + AppImage（linux，rpm 需系统 rpmbuild）
-make core-dist    # 构建核心 deb + rpm + tar.gz（服务器版，需 nfpm）
-make gui-win      # 构建 Windows nsis 安装包（本机 mingw + wine）
-make gui-win-docker  # 构建 Windows nsis 安装包（全 docker 自足，含 wine）
-make gui-dist-dir # 仅解包产物（调试用，不产安装包）
-```
-
-产物在 `cmd/build/gui-dist/`：`ModelRouter-<ver>.AppImage`、
-`modelrouter-gui_<ver>_amd64.deb`、`ModelRouter Setup <ver>.exe`（win）；
-核心包在 `cmd/build/dist/`：`llmsproxy_<ver>_amd64.deb`、`llmsproxy-<ver>.x86_64.rpm`。
-rpm 需系统 `rpmbuild`。每次打包后都会跑 `packaging/verify-dist.sh` 产物体检
-（大小下限门禁——1.3.0 曾因宿主 wine 损坏产出 264KB 的残缺 exe 却没有被拦住）。
-发布时会把产物重命名为 `ModelRouter-Desktop-*`（GUI）与 `ModelRouter-Headless-*`（纯后端）
-两类上传到 GitCode Release。
-
-#### Windows 打包（docker 固化方案，推荐）
-
-`make gui-win-docker` 一步完成：交叉编译核心 + NSIS 打包**全在 docker 内**，宿主机
-**无需 mingw、无需 wine、无需 node**。镜像内置 wine（含 i386 运行时），修复了
-此前宿主 wine 损坏（缺 syswow64 → `c0000135`）导致产出 264KB 残缺 exe 的问题。
-
-- 构建镜像：`cmd/gui/docker/win-builder/Dockerfile`（`golang:1.25` + mingw-w64 +
-  node + wine32/wine64 + 预编译 LuaJIT windows 版 `lua51.dll` 与 import lib）。
-  LuaJIT 源码取自 gitcode 镜像（国内可达），github 兜底；Go 依赖走 `https://goproxy.cn`。
-- 一键发布：`cmd/gui/scripts/dist-win-docker.sh` —— 核心编译 + electron-builder
-  NSIS 全部在容器内执行（`USE_SYSTEM_WINE=true` 走镜像内的 wine）；electron-builder
-  的 ~400MB 工具链缓存挂到 docker volume，二次构建增量复用。
-- 首次运行自动构建 `modelrouter/win-builder` 镜像（约 5–10 分钟）；已存在则直接复用。
-- 打包完成后自动跑 `packaging/verify-dist.sh` 门禁：exe 低于 5MB / deb·rpm·7z 低于
-  10MB 即失败——残缺安装包不可能再溜进 Release。
-
-```bash
-# 全 Windows 发布（核心重编 + NSIS）：
-make gui-win-docker
-```
-
-### 特性
-
-- **内嵌核心**：自动拉起捆绑的 `llmsproxy`（LuaJIT 版），配置/密钥/adapter 存于
-  用户数据目录 `<userData>/profile/`，首次运行自动生成随机 admin key 写入 `keys`
-  （非 seed）并自注入，WebUI 免登录、不会提示"更换初始密钥"。端口可在设置中更改（默认 8787）。
-- **系统托盘**：内核状态点、开机自启/静默启动开关、启停/重启内核、退出。
-- **静默启动**：`--silent` 参数或设置项，启动仅驻托盘不弹窗；首次运行始终显示窗口。
-- **开机自启**：Windows/macOS 用 `setLoginItemSettings`；Linux 写
-  `~/.config/autostart/modelrouter-gui.desktop`，静默开启时自动附加 `--silent`。
-
-## 目录
-
-```
-cmd/llmsproxy            # 入口（纯 Go 服务端）
-cmd/gui                  # Electron 桌面 GUI（内嵌核心 + 托盘 + 打包）
-internal/config          # YAML 配置加载/校验
-internal/lua             # LuaJIT VM + worker 池 + AdapterCache + 内置适配器 (embed)
+旧路径与旧环境变量在读取侧仍被接受，因此升级不会因为顺序问题立刻失效。
+`deploy.sh` 在检测到旧 `/etc/llmsproxy` 而新配置不存在时会**中止**，避免部署出
+一个「能启动但没有 key / 没有源」的空实例。
+
+插件 SDK 同名调整：唯一公开入口是 `modelrouter`（`llmsproxy` 仅作兼容别名），
+详见 [docs/plugins.md](docs/plugins.md)。
+
+## 目录结构
+
+```text
+cmd/modelrouter          # 入口（纯 Go 服务端）
+cmd/gui                  # Electron 桌面壳（内嵌同一份核心）
+internal/config          # 配置解析、加密存储、运行时文件
+internal/core            # 调度、内核动作、状态与审计
+internal/lua             # LuaJIT VM + 弹性 worker 池 + 内置适配器 (embed)
+internal/lua/plugins     # 可选插件示例（不随二进制/发行包分发）
 internal/provider        # Provider(HTTP) + Registry(路由)
-internal/gateway         # OpenAI 兼容 HTTP 服务 + 鉴权 + SDK/流式 + WebUI
+internal/gateway         # OpenAI 兼容 HTTP 服务 + 鉴权 + 流式 + WebUI
 internal/types           # 统一格式 & OpenAI wire 类型
+packaging/               # deb/rpm/tar 打包、systemd unit、迁移脚本
+docs/                    # 主题文档
 ```
 
-## 开发协作
-
-- **Git 分支工作流**：主分支 / 特性分支 / 发布分支的规范见
-  **[docs/git-workflow.md](docs/git-workflow.md)**（[English](docs/git-workflow-en.md)）。
-  主线：`main` 永可部署 → 特性合入 `main` → 切 `release/vX.Y.Z` 打 tag 发布 →
-  hotfix 提交发布分支并 cherry-pick 回 `main`。
-- **Lua 适配器协议**：见 **[docs/lua-adapters.md](docs/lua-adapters.md)**
-  （[English](docs/lua-adapters-en.md)）。
-
-## 测试
+## 开发与测试
 
 ```bash
-go test -tags luajit ./...
+go test -tags luajit ./...          # 全量
+go test -tags luajit ./internal/lua # 单包（LuaJIT 相关改动必跑）
 ```
 
-覆盖：配置校验、适配器加载/变换、签名钩子、Gateway 鉴权、SDK 结算、SSE 流式、
-模型路由、多模态透传与 thinking 预算。
+- 分支与发布纪律见 [docs/git-workflow.md](docs/git-workflow.md)。
+- 插件放置在 `plugin_dir`，网关启动时**不会**创建、覆盖或删除任何插件文件；
+  升级桌面端也不会动它们。

README_EN.md

@@ -2,625 +2,148 @@
 
 > **中文**: [README.md](./README.md)
 >
-> **AI-assisted**: This project is developed with AI-assisted coding (code and docs are produced in collaboration with AI, reviewed by humans).
+> **AI-assisted**: code and documentation were produced with AI assistance and
+> reviewed by a human.
 
-A lightweight, unified OpenAI-compatible LLM gateway for internal networks —
-configure once, and let multiple services (agents, SDKs, bots) share many
-upstreams (DeepSeek, Qijiar, OpenAI, Anthropic, Gemini, Groq, Mistral, Ollama,
-KimiCode…) behind a single endpoint. Pick a specific model or use **AUTO** mode,
-which routes to the best healthy upstream by configured priority.
+An intranet gateway for people with several LLM subscriptions: configure once and
+let multiple services share the upstream capacity. Pick a model explicitly, or let
+`AUTO` choose by priority among the sources that are currently healthy.
 
-> **Lightweight, no recompile to add sources**: the gateway is a single Go
-> binary (8-12 MB, ~8 MB stripped; zero runtime dependencies). Adding or switching an upstream is
-> just a `sources` entry in `config.yaml` (or via the WebUI) or a `.lua`
-> adapter — **no Go changes, no recompile**. Changes submitted through the
-> WebUI take effect immediately (hot reload); editing `config.yaml` or a `.lua`
-> file under `adapter_dir` by hand requires a process restart.
+The gateway itself is a single Go binary (~10 MB, no runtime dependencies). Adding
+or switching an upstream is a `sources` entry in `config.yaml` (or the WebUI), or a
+dropped-in `.lua` adapter — **no Go changes, no rebuild**. Changes submitted through
+the WebUI take effect immediately; editing `config.yaml` or a `.lua` adapter file
+directly requires a restart.
 
-## Overview
+## What it does
 
-Exposes a standard `OpenAI Chat Completions` API (`/v1/chat/completions` +
-`/v1/models`) to your internal network, translating between multiple upstream
-protocols via **Lua adapters**. Supports one-shot calls, SSE streaming,
-**AUTO model routing**, multimodal passthrough, and tool calls.
-
-Extracted and independently evolved from the multi-source LLM adapter layer of
-[HomeAgent](https://gitcode.com/JianFeeeee/HomeAgent)
-(`internal/agent/api/provider.go` + `internal/lua/adapters/*`).
-
-## Features
-
-- **Multi-source**: any number of upstream sources in one process, routed by the
-  request's `model`.
-- **AUTO mode**: with `default_model: AUTO`, scheduling follows the tiered AUTO
-  chain saved on the Priority page (see below).
-- **Unified output**: every source speaks OpenAI format (incl.
-  `reasoning_content`, `tool_calls`, `usage`).
-- **Image generation**: `POST /v1/images/generations`, routed to models with
-  `kind: image`.
-- **Per-model quota**: each key gives every model its own token and request caps
-  plus a reset period (hour/week/month/custom N hours). One model running out
-  affects only that model — the key's other models keep working. Exhaustion
-  answers 429 + `Retry-After` naming the model, so a client resumes when the
-  window rolls over; admin keys are never capped.
-- **Multimodal**: `content` arrays (`image_url` etc.) pass through losslessly;
-  Anthropic/Gemini/Ollama are translated automatically.
-- **LuaJIT VM**: golua-binding LuaJIT; each adapter has its own VM + worker
-  pool for safe concurrency. Pools are **elastic**: the sum of `max_concurrent`
-  over an adapter's sources is a ceiling, not a preallocation — states are booted
-  on demand and reclaimed when demand drops, so an idle gateway holds close to
-  zero Lua states. The grow step follows the adapter's max concurrency
-  (`clamp(ceil(max/8), 1, 8)`, and never exceeds the number of queued callers);
-  the shrink step follows the live connection count
-  (`clamp(ceil(slack/(1+in_use)), 1, slack)`), so an idle pool collapses in one
-  round while a busy one gives up a single state at a time.
-- **On-demand request logs**: dashboard totals are computed from the **full**
-  audit history (streamed once at startup and released — 29 MB / 221k lines in
-  ~260 ms), while the raw records are never held in memory: the dashboard loads
-  one screen, scrolling pages the rest straight off disk, CSV export streams in
-  O(1) memory, and leaving the page releases everything.
-- **Self-healing cooldown**: cooldown is capped at 5 minutes and, past the
-  window's midpoint, exactly one probe request is allowed through; a recovered
-  upstream (or a reset quota) returns to full rotation on that probe instead of
-  waiting out the window.
-- **disable_thinking**: `disable_thinking: true` toggles reasoning per-request.
-- **Lua adapter protocol**: each source mounts a `.lua` adapter with
-  `transform_request` / `transform_response` / `transform_stream_chunk` — all
-  protocol differences live in Lua, Go only schedules and proxies.
-- **Signature / header hooks**: adapters may define `build_headers(meta)` to
-  inject/sign request headers before the HTTP call (e.g. KimiCode style
-  app-validation), with helpers like `hmac_sha256_hex`, `sha256_hex`,
-  `base64_encode`.
-- **WebUI**: built-in management page to view/add/edit sources & models, edit
-  the AUTO chain, and manage per-key model scopes, persisted to a runtime file.
-- **Encrypted secrets at rest**: upstream `api_key`, custom header values and
-  gateway keys are stored AES-256-GCM encrypted (`master.key` 0600 next to the
-  runtime file, or `LLMS_PROXY_MASTER_KEY`).
-- **Live source probing**: the status page probes each source
-  (`GET {base}/models`, falling back to a minimal chat call) without touching
-  the scheduler's backoff state; errors are shown on hover.
-- **Auth**: the gateway validates client keys via `gateway_keys`; independent
-  from each upstream's own key.
-- **Streaming**: SSE `chat.completion.chunk` with a role-first chunk and
-  `[DONE]` terminator.
+- **One OpenAI-compatible API**: `/v1/chat/completions`, `/v1/models`,
+  `/v1/images/generations`. Every source speaks the same shape outwards, including
+  `reasoning_content`, `tool_calls` and `usage`.
+- **Many sources, one AUTO chain**: configure any number of upstreams; `AUTO`
+  follows the priority tiers saved in the WebUI, fails over between them, cools
+  down broken sources and probes them back to health at half-cool.
+- **Protocol differences live in Lua**: each source binds a `.lua` adapter that
+  transforms requests, responses and stream chunks both ways, and can sign headers
+  via `build_headers(meta)`. Vendor knowledge stays out of the Go kernel.
+- **Plugins are independent**: business extensions (channel login, billing, quota
+  probing) are external Lua plugins loaded from `plugin_dir`. They are not compiled
+  into the binary and not shipped in release archives, and they talk to the kernel
+  only through the `modelrouter` SDK.
+- **Multi-tenant keys**: gateway keys are separate from upstream keys, and can be
+  scoped to a model set, an AUTO chain and a quota.
+- **Encrypted at rest**: upstream keys, custom header values and gateway keys are
+  sealed with AES-256-GCM; `master.key` is 0600, or injected via
+  `MODELROUTER_MASTER_KEY`.
+- **Observability**: live source probing, HTTP access log, config-change audit,
+  request and usage CSV export. Writes are kept off the request hot path.
 
 ## Quick start
 
 ```bash
-GOMODCACHE=... GOPROXY=off go build -tags luajit -o llmsproxy ./cmd/llmsproxy
-./llmsproxy -config config.yaml    # first run generates a default config and prints a random admin key
+# LuaJIT development libraries are required; -tags luajit is mandatory
+GOMODCACHE=... GOPROXY=off go build -tags luajit -o modelrouter ./cmd/modelrouter
+
+# First run generates a default config and prints a random admin key
+./modelrouter -config config.yaml
 ```
 
-> Depends on [golua](https://github.com/aarzilli/golua) (LuaJIT bindings).
-> You **must** build with `-tags luajit`; otherwise the built-in superset
-> gopher-lua path is used (behavior differs slightly).
+> Depends on [golua](https://github.com/aarzilli/golua) (LuaJIT bindings). Without
+> `-tags luajit` the build falls back to gopher-lua, which behaves slightly
+> differently.
 
-> **No config is shipped in the repo** (config files carry real keys). On
-> first run the binary generates a default config at the `-config` path: a
-> random admin key (printed to the startup log), loopback-only `127.0.0.1:8080`,
-> and a keyless source ready to chat. Rotate the admin key in the
-> WebUI after first login.
+> **No config in the repo**: the repository ships no config file (they contain
+> keys). The generated default binds `127.0.0.1:8080` only; rotate the admin key in
+> the WebUI after the first login.
 
 ```bash
-# no key -> 401
+# No key -> 401
 curl http://127.0.0.1:8080/v1/models
 
-# one-shot ($KEY = the admin key printed at first startup)
+# Non-streaming ($KEY = the admin key printed at first start)
 curl -H "Authorization: Bearer $KEY" \
      -d '{"model":"AUTO","messages":[{"role":"user","content":"hi"}]}' \
      http://127.0.0.1:8080/v1/chat/completions
 
-# streaming
+# Streaming
 curl -N -H "Authorization: Bearer $KEY" \
      -d '{"model":"AUTO","stream":true,"messages":[{"role":"user","content":"hi"}]}' \
      http://127.0.0.1:8080/v1/chat/completions
 ```
 
-Any OpenAI SDK works: point `base_url` at the gateway, use any authorized
-gateway key as the API key.
-
-## Configuration
-
-Generated on first run at the `-config` path (default `./config.yaml`). Core fields:
-
-```yaml
-listen: 127.0.0.1:8080        # bind address (keep internal/loopback)
-gateway_keys: [sk-gw-<random>] # initial admin key seed (random per install), migrated to the store on first start
-default_model: AUTO           # when model is unroutable, follow the AUTO chain
-adapter_dir: adapters         # Lua adapter dir; created+seeded if missing, read-only otherwise
-runtime_file: runtime.json    # WebUI-edited sources/keys/AUTO chain persist here
-
-sources:
-  - name: deepseek
-    base_url: https://api.deepseek.com
-    api_key: sk-...            # or api_key_env: SOME_ENV to read from an env var
-    adapter: deepseek
-    max_concurrent: 8
-    models:
-      - id: deepseek-v4-flash
-        priority: 100          # YAML sources seed the AUTO chain on first start
-        kind: chat
-      - id: deepseek-v4-pro
-        priority: 60
-        kind: chat
-    # static headers (take precedence over adapter defaults)
-    headers: { X-Tenant: prod }
-    # passthrough metadata for the Lua build_headers hook
-    meta: { app_id: x, app_secret: y }
-    temperature: 0.7
-    max_tokens: 4096
-    timeout: 120s              # request timeout, default 120s
-```
-
-Sensitive fields in the runtime file (`runtime_file`) are encrypted at rest:
-
-- AES-256-GCM, stored as `enc:v1:<base64>`.
-- Master key source: env var `LLMS_PROXY_MASTER_KEY` (64 hex chars), else a
-  `master.key` file next to the runtime file; if neither exists it is generated
-  on first start (mode 0600).
-- Old plaintext files still load; the first UI save migrates the whole file to
-  ciphertext.
-- Back up `master.key` with your config — losing it makes the secrets
-  undecryptable. Do not commit it.
-
-### Gateway keys (multi-key)
-
-Gateway auth uses a "multi-key + role + model scope" model. Keys are persisted
-under the `keys` field of the runtime file (encrypted at rest):
-
-- The `gateway_keys` config is only an **initial admin key seed** — it is
-  migrated into the runtime store on first start and no longer drives auth.
-- **Important: after first start, replace the admin key via the WebUI Keys
-  page.** The seed key is written in plaintext in `config.yaml`, so keeping it
-  active is a security risk; create a new admin key, log in with it, then
-  delete the seed key.
-- The WebUI **Keys page** creates/deletes keys. Each key has a role (`admin`
-  manages everything, `user` sees only its own key) and an optional **model
-  scope** (model + source + token quota + reset period). Each model brick shows
-  its own budget badge (e.g. `1.0K 77×·1h`, `∞` when uncapped); clicking it
-  edits that model's token quota, request quota and reset period. Quotas apply
-  per model, so one model running out never blocks the key's others. The
-  "My key" view shows a user its model scopes and their budgets.
-- Clients authenticate with any authorized key's plaintext as
-  `Authorization: Bearer <key>`.
-- Deleting a key removes it from the store immediately.
-
-#### Per-key usage quota
-
-**Quotas are per model.** Each key's `models[]` list gives every model its own
-token budget and request budget. One model running out affects only that model
-— the key's other models keep working.
-
-```yaml
-keys:
-  - key: sk-gw-<hex>
-    role: user
-    name: agent-alice
-    models:
-      - model: deepseek-v4-flash
-        token_quota: 1000000   # this key's token budget for this model
-        req_quota: 20000       # requests within the window
-        period: nhour          # "" | hour | week | month | nhour
-        hours: 6               # n-hour only
-      - model: AUTO
-        token_quota: 5000000   # AUTO is a quota entry like any other
-        period: hour
-      - model: kimi-k3        # no quota listed = unlimited
-```
-
-- **There is deliberately no key-wide total.** A key-wide cap would make
-  "move A's budget to B" a global reallocation; per-model budgets keep each
-  model independently controllable, so it stays visible which model is
-  actually consuming the spend.
-- `period`: empty = never resets (lifetime total); `hour` / `week` / `month` =
-  fixed windows; `nhour` + `hours` = a custom hour count. **A misspelled
-  period is rejected at write time** rather than silently becoming a
-  never-resetting quota.
-- Quotas are isolated per key *and* per model within a key: one key exhausting
-  `m1` never draws on another key's budget, and never blocks the same key's
-  `m2`.
-- Usage counts chat, streaming and image requests, and survives a restart by
-  replaying the audit log (40 days retained, covering the longest `month`
-  window).
-- An exhausted quota returns **429 + `Retry-After`**
-  (`rate_limit_exceeded`) and the message names the model that ran out, so a
-  client resumes when the window rolls over; a model the key may not use stays
-  403. **Admin keys are never capped** (quotas on their scopes are not
-  enforced either), so a cap can never lock the operator out.
-- Buckets are whole unix hours, so a window frees up at most an hour late
-  (deliberately freeing late rather than overspending).
-- `PUT /api/keys/{key}` submits quotas by submitting `models` — the caps are
-  part of the scope, so there is no second budget that can drift out of sync
-  with the model list. An explicit `0` lifts that model's cap.
-
-##### Quota rejection vs capacity rejection
-
-A quota is checked at the entry point and occupies no upstream slot, so it is a
-cheap rejection. Running out of capacity instead waits out the bounded
-`busyWait` before answering 503. Measured (capacity 8, 0.6 s/request, 100
-concurrent clients):
-
-| Case | Status | Latency | Reaches upstream |
-|---|---|---|---|
-| Quota spent (`req_quota` 50, 100 concurrent) | `429 rate_limit_exceeded` + `Retry-After` | **19–21 ms** | no |
-| Capacity exhausted (same topology, no quota) | `503 upstream_error` | ~2.6 s (`busyWait`) | overflow: no |
-
-So a client can tell "wait for the window to roll over" apart from "wait for
-upstream capacity" — the first only needs patience, the second usually means
-back off concurrency or switch sources.
-
-##### What the quota costs
-
-Buckets are kept per (key, model, whole unix hour) for 40 days. Measured on an
-AMD 7840HS:
-
-- A window query scans the window, not the whole retention: 49 ns for 1 h,
-  55 ns for 24 h, 3.9 µs for 30 d.
-- Recording one request: 283 ns, 3 allocations (unchanged from before this
-  feature; the allocations come from the record ring buffer).
-- Memory: the production shape (7 keys x 8 models x 2 sources at full 40-day
-  retention) costs about **3.7 MB**. The `source::model` bucket used by a
-  source-pinned quota is created **lazily** — it is maintained only once some
-  quota actually pins that source, because writing it on every record would
-  cost an extra 18 MB at 20 keys x 8 models x 3 sources.
-  The trade-off: usage recorded before a pinned quota was configured cannot be
-  split by source afterwards, so **a pinned quota may under-count its first
-  window**.
-
-```
-HTTP/1.1 429 Too Many Requests
-Retry-After: 2100
-{"error":{"type":"rate_limit_exceeded","message":"token quota exceeded for \"deepseek-v4-flash\" (5000000/5000000)"}}
-```
-
-### Model routing
-
-`/v1/chat/completions` `model` resolution order:
-
-1. `source/model` or `source:model` prefix → pinned source;
-2. exact match of a source's `model`;
-3. with `default_model: AUTO` → follow the AUTO chain from the Priority page
-   (see below);
-4. otherwise the request errors.
-
-Any OpenAI client can pin to a source by setting `model` to its
-`name/anything`; `AUTO` (or the gateway's `default_model: AUTO`) follows the
-tiered AUTO chain.
-
-### AUTO chain (Priority page)
-
-AUTO scheduling is driven **only** by the rules saved on the Priority page. The
-numeric `models[].priority` in source configs no longer participates in
-scheduling and is no longer shown. **Chat and image generation are two
-independent chains**, switched in the Priority page via the `Chat / Image` toggle:
-
-- **Chat chain** (`auto` field): serves `POST /v1/chat/completions` with
-  `model: AUTO`; only `kind: chat` models are honored (image slots are skipped).
-- **Image chain** (`auto_image` field): serves `POST /v1/images/generations` with
-  `model: AUTO`; only `kind: image` models are honored. With no image chain
-  configured, AUTO image generation falls back to legacy discovery (all sources
-  exposing an image model, in registry order) for backward compatibility.
-
-Each rule is one "slot": `{ model, source, tier, token_quota, period, hours }`.
-
-- `tier` is the priority tier: models in the same tier sit side by side and
-  share it; tiers run high → low.
-- The same model may appear in several slots (e.g. A low → B low → A high →
-  B high) and is tried in tier order.
-- `token_quota` > 0 means the slot is skipped once its tokens within the reset
-  window are exhausted; `period` supports `hour` / `week` / `month` / `nhour`
-  (with `hours`); empty = unlimited.
-- Within a tier, slots run by preference score; cooling / quota-exhausted /
-  hard-failed slots fall through to the next tier, and total failure returns
-  503 with a per-tier summary.
-
-### Sources vs. adapters
-
-- A **source** describes a connection to an upstream: `name`, `base_url`,
-  `api_key`, model list and priorities.
-- An **adapter** is the protocol translation logic (Lua script): converts the
-  unified OpenAI request to the upstream's native format and back.
-- One adapter serves many sources (e.g. `openai.lua` for any OpenAI-compatible
-  site); one source can switch adapters via its `adapter` field.
-- Sources decide *whom to talk to and which models to expose*; adapters decide
-  *how to talk*. They are linked by the `adapter` field inside each `sources[]`
-  entry.
-
-Loading flow (assembled by `Core` in `internal/core`):
-
-1. On startup `lua.NewVM(adapter_dir)` loads all adapters: built-in ones
-   (embedded at compile time) + same-name override files under `adapter_dir`.
-2. `config.Load` reads `config.yaml`; `config.NewStore(runtime_file)` reads
-   WebUI-edited runtime sources; both are merged by name.
-3. `rebuildRegistry` creates a `provider.Provider` per source (holding its
-   adapter), sizing the adapter's worker pool from the source's concurrency
-   limit.
-4. Incoming requests are routed by `Registry` to a provider, which calls the
-   adapter's `transform_request` → HTTP call → `transform_response` /
-   `transform_stream_chunk`.
-
-"Add/Edit source", "Upload adapter", "AUTO chain edits" and "per-key model
-scopes" in the WebUI take effect immediately (written to the runtime file /
-`adapter_dir`, then reassembled — no restart). Directly editing `config.yaml`
-or files under `adapter_dir` requires a process restart.
-
-Deleting a source/adapter in the WebUI truly removes runtime (UI-created)
-sources and `.lua` files; base sources defined in `config.yaml` cannot rewrite
-that file, so they are hidden via a deletion tombstone (still hidden after
-restart) and can be restored by re-adding the same name in the UI.
-
-### Source templates (multi-key balancing)
-
-When several upstream keys share one config (URL / adapter / model list /
-concurrency / rpm), copying the whole source block N times is wasteful. A
-**source template** stores every source field except `name` and `api_key`, so you
-create N key-only-different sources from one shared skeleton — natural multi-key
-load balancing (each expanded source schedules, cools down, and reports health
-independently).
-
-- WebUI Sources page, top-right **Templates** button: list all templates with
-  edit / delete / create.
-- Add-source dialog has two header buttons:
-  - **From template** → pick a template → form auto-fills (name + key still
-    yours to enter);
-  - **As template** → save the current form's non-key/name fields as a template.
-- Templates persist in the runtime file's `source_templates` field alongside
-  runtime sources (hot-managed, apply immediately). Templates hold no secret
-  keys, so they're safe to share / version-control.
-
-A template is just a recipe — it doesn't become a source by itself. Only the
-key-bearing sources created from it carry real traffic.
-
-### Memory footprint (measured and tuned)
-
-Memory is not a constant: it scales with the **number of configured sources**
-(each source owns an `http.Transport` with its connection pool plus per-model
-health state), not with uptime. Measured on this machine (Linux x86_64, 12 cores):
-
-| Deployment shape | Startup RSS | Settled RSS |
-|---|---|---|
-| 1 source / 1 adapter (minimal) | ~4 MB | ~10 MB |
-| 1 source + a 29 MB audit history | ~19 MB | ~20 MB |
-| **16 sources / 13 adapters / 59 models (this host)** | ~28 MB | **~32-35 MB** |
-
-> For reference, the same production config used **~105 MB** before this round of
-> work. The reduction comes from three places: raw audit records are no longer kept
-> resident (~25 MB — the aggregates still scan the full history, but the scan
-> releases as it goes), Lua state pools no longer grow monotonically, and the two
-> runtime knobs below.
-
-Breakdown of the production instance (per-region, from `/proc/<pid>/smaps`):
-
-| Region | RSS | Notes |
-|---|---|---|
-| Go heap | ~14 MB | provider/registry/scheduler state + connection-pool buffers |
-| Other anonymous (thread stacks / LuaJIT chunks / runtime) | ~12 MB | scales with thread count and loaded adapters |
-| Binary text+rodata | ~8 MB | mapped executable pages (read-only, reclaimable by the kernel) |
-| Go runtime reservations | ~4 MB | the 2 GB+ you see in VSZ is address space, not physical memory |
-| Shared libraries | ~3 MB | libc / libluajit / libm |
-
-**Two recommended deployment knobs** (environment only, no code change):
-
-```ini
-# /etc/systemd/system/llmsproxy.service
-Environment=GOGC=50
-Environment=MALLOC_ARENA_MAX=2
-```
-
-- `MALLOC_ARENA_MAX=2`: LuaJIT allocates through cgo into glibc malloc, and glibc
-  allows up to `8 x nproc` per-thread arenas. Every OS thread that touches malloc
-  claims one (~1 MB each) and **never returns it to the OS**. Measured: 8-15
-  arenas (7-12 MB) down to **0**.
-- `GOGC=50`: halves the Go heap growth target. **It does nothing on its own** —
-  the heap it saves is immediately consumed by additional glibc arenas (measured
-  20.3 -> 21.5 MB, i.e. slightly worse) — so it must be paired with
-  `MALLOC_ARENA_MAX`. Together they cut settled RSS by **~19%**. The gateway is
-  I/O bound (1min10s of CPU per 9 hours here), so the extra GC cycles are free.
-
-> The desktop build (Electron) already injects both when spawning the embedded
-> core (`cmd/gui/main.js`); exporting either variable yourself overrides it. If you
-> write your own systemd unit or container spec, add them there.
-
-Optional: `GOMEMLIMIT=48MiB` as a soft ceiling saves roughly 1 MB more, at the
-cost of GC turning aggressive as the limit approaches. Not worth it for a relaxed
-deployment.
-
-### disable_thinking
-
-With `"disable_thinking": true` in the request body, the gateway passes it to
-each adapter; the DeepSeek adapter maps it to `extra_body.thinking.type =
-"disabled"`, other sources follow their own protocol.
-
-### WebUI
-
-Built-in admin page at `GET /`; after login you can:
-
-- **Status**: source online state (live probe, error on hover), per
-  model/source/key usage stats, request records filtered by key, and CSV export
-  over a time range; click a model to pin the connection snippet to it.
-- **Chat**: streaming / non-streaming debug.
-- **Keys**: create/edit gateway keys, set per-key model scopes (model + source +
-  token quota + reset period); admins manage all keys, users see only their own.
-- **Priority**: drag blocks to build the tiered AUTO chain.
-- **Sources**: add/edit/delete upstream sources online (secrets stored
-  encrypted).
-- **Adapters**: upload / delete Lua adapter scripts.
-
-Changes persist to `runtime_file` (survive restarts).
-
-## Lua adapter protocol
-
-Full API: **[Lua adapter docs](docs/lua-adapters.md)**
-([中文](docs/lua-adapters-en.md)).
-
-Each adapter is a Lua script returning a table
-(`internal/lua/adapters/<name>.lua`), overridable by a same-name file under
-`adapter_dir` — **no recompile needed**, restart to take effect; uploads via
-the WebUI take effect immediately.
-
-```lua
-return {
-  name = "mysrc",
-  version = "1.0.0",
-  endpoint = "/chat/completions",        -- upstream path (source.endpoint overrides)
-  headers = { ["X-Static"] = "v" },      -- static default headers (fallback)
-
-  -- request: convert unified OpenAI req -> upstream native format, return string
-  transform_request = function(raw_json) ... end,
-
-  -- response: convert upstream response to unified format string
-  --   { content, reasoning_content, finish_reason, token_usage{...}, tool_calls[{...}] }
-  transform_response = function(raw_json) ... end,
-
-  -- streaming chunk: convert upstream SSE data to { content, done, ... }; "" skips
-  transform_stream_chunk = function(raw_chunk) ... end,
-
-  -- [optional] dynamic headers / signature hook
-  -- meta = { url, method, body, api_key, timestamp, source={ name, meta={...} } }
-  build_headers = function(meta) return { ["X-App-Sign"] = sign } end,
-}
-```
-
-Shared helpers: `hmac_sha256_hex(key, data)`, `sha256_hex(data)`,
-`base64_encode(s)`, `tohex(s)`, `json.encode/decode`, `log(level, msg)`.
-
-### Built-in adapters
-
-`openai` `deepseek` `anthropic` `gemini` `github` `groq` `mistral` `ollama`
-`kimicode` `qwen38` `sensenova` `trae`.
-
-`anthropic`/`gemini`/`ollama` include multimodal conversion
-(`image_url` → their native format); with `disable_thinking` the `deepseek`
-adapter sets `extra_body.thinking.type` to `disabled`.
-
-**kimicode** demonstrates `build_headers`: the cloud validates the calling
-app, so you HMAC-sign timestamp+URL+body with `meta.app_secret` and add
-`X-App-Sign`-style headers. Configure `sources[].meta.{app_id, app_secret,
-app_agent}`.
-
-## Desktop GUI (Electron)
-
-Optional standalone product for desktop users on your internal network: an
-**embedded ModelRouter core** (Clash Verge-style): one-click start/stop, system
-tray, autostart, silent launch, and a pixel-for-pixel embedded full WebUI
-(Status/Chat/Keys/Priority/Sources/Adapters — no login needed). Server users
-keep using the plain Go binary.
-
-### Two installer kinds: Headless and Desktop
-
-Release installers ship in two flavors for different audiences:
-
-- **Headless (server)**: the plain core binary (no Electron), config-driven,
-  for servers / containers / systemd / unattended runs. Ships as `deb` + `rpm`
-  (systemd unit with memory tuning + example config) and `tar.gz` (binary +
-  `config.example.yaml` + `adapters/` + README), Windows as `.zip`.
-- **Desktop (desktop app)**: the Electron GUI installer with an embedded core,
-  for personal daily desktop use.
-
-### GUI vs. plain backend
-
-| Scenario | Pick | Why |
-| ---- | ---- | ---- |
-| Server / intranet gateway / unattended long-running | **Headless** (single binary) | ~8 MB binary; RSS scales with the number of sources (~10 MB for one, ~32-35 MB for 16 — see [Memory footprint](#memory-footprint-measured-and-tuned)), zero-dep single process — drop it into systemd or any container, remote admin |
-| Personal desktop daily use / multi-device intranet sharing | **Desktop** (GUI) | no-login embedded WebUI, tray one-click, autostart, silent background — for non-CLI users |
-| Windows desktop | **Desktop** | plain backend needs manual service registration; GUI ships native tray/autostart |
-| CI one-shot 3-platform installers | **Desktop packaging scripts** | `make gui-*` emits deb / AppImage / NSIS, drops straight into a release pipeline |
-
-Both share the exact same `llmsproxy` core (LuaJIT build) — configs and
-adapters are fully compatible, freely interchangeable.
-
-### Build & run
+Any OpenAI SDK works: point `base_url` at the gateway and use an authorized gateway
+key as `api_key`.
+
+## Documentation
+
+The detailed documentation is currently written in Chinese.
+
+| Topic | Document |
+|---|---|
+| Implementation and the full feature list | [docs/features.md](docs/features.md) |
+| Configuration: `config.yaml`, model discovery, multiple keys | [docs/configuration.md](docs/configuration.md) |
+| Routing: model routing, AUTO chain, cooldown, sources vs adapters, source templates | [docs/routing.md](docs/routing.md) |
+| Adapters: Lua protocol and built-in adapters | [docs/adapters.md](docs/adapters.md) · [docs/lua-adapters-en.md](docs/lua-adapters-en.md) (English) |
+| Plugins: SDK contract, hooks, UI extensions, HTTP endpoints | [docs/plugins.md](docs/plugins.md) |
+| Desktop app: Headless / Desktop artifacts, build and run | [docs/gui.md](docs/gui.md) |
+| Performance: measured memory use, tuning, thinking budget | [docs/performance.md](docs/performance.md) |
+| WebUI usage | [docs/webui.md](docs/webui.md) |
+| Development: layout, tests, branch and release discipline | [docs/development.md](docs/development.md) · [docs/git-workflow-en.md](docs/git-workflow-en.md) (English) |
+| Deployment and migration | [docs/deployment.md](docs/deployment.md) |
+
+## Renamed from llmsproxy
+
+The product name has been ModelRouter for a while; the binary, config directory,
+service name and environment variables are now `modelrouter` as well:
+
+| Before | After |
+|---|---|
+| `llmsproxy` (binary) | `modelrouter` |
+| `/etc/llmsproxy/` | `/etc/modelrouter/` |
+| `llmsproxy.service` | `modelrouter.service` |
+| `LLMSPROXY_*` | `MODELROUTER_*` |
+
+Existing installs migrate with a dry-run-by-default script (copies, never moves,
+and can roll back):
 
 ```bash
-cd cmd/gui
-npm install
-make gui            # or npm run dev — dev run (needs make build first for bin/llmsproxy)
-make gui-deb        # local deb (share with other Linux users)
-make gui-dist       # deb + rpm + AppImage (linux; rpm needs system rpmbuild)
-make core-dist      # core deb + rpm + tar.gz (server edition; needs nfpm)
-make gui-win        # Windows NSIS (host mingw + wine)
-make gui-win-docker # Windows NSIS, fully dockerized (wine in image; no host mingw/node/wine)
+sudo ./packaging/migrate-to-modelrouter.sh              # preview
+sudo ./packaging/migrate-to-modelrouter.sh --apply      # apply
+sudo ./packaging/migrate-to-modelrouter.sh --rollback --apply
 ```
 
-Artifacts land in `cmd/build/gui-dist/`: `ModelRouter-<ver>.AppImage`,
-`modelrouter-gui_<ver>_amd64.deb`, `ModelRouter Setup <ver>.exe` (win); core
-packages land in `cmd/build/dist/`: `llmsproxy_<ver>_amd64.deb`,
-`llmsproxy-<ver>.x86_64.rpm`. Both target families run
-`packaging/verify-dist.sh` afterwards — a size floor that rejects degenerate
-installers (1.3.0 shipped a 264 KB NSIS stub because the host wine was broken
-and nothing caught it). At release time artifacts are renamed into the
-`ModelRouter-Desktop-*` (GUI) and `ModelRouter-Headless-*` (plain backend)
-families before upload to GitCode Releases.
+Old paths and old environment variable names are still accepted on the read side,
+so an upgrade cannot fail merely because of ordering. `deploy.sh` refuses to run
+when it finds the old `/etc/llmsproxy` without a new config, because that would
+deploy a gateway with no key and no sources.
 
-#### Windows packaging (dockerized, recommended)
-
-`make gui-win-docker` does cross-compile + NSIS **entirely in docker**; the host
-needs **no mingw, no wine, no node**. Wine (with the i386 runtime) lives inside
- the image, fixing the broken-host-wine failure that produced the 264 KB stub
-installer.
-
-- Image: `cmd/gui/docker/win-builder/Dockerfile` (`golang:1.25` + mingw-w64 +
-  node + wine32/wine64 + prebuilt LuaJIT for Windows: `lua51.dll` + import lib).
-  LuaJIT source comes from the gitcode mirror (reachable in CN) with github
-  fallback; Go modules go through `https://goproxy.cn`.
-- One-shot release: `cmd/gui/scripts/dist-win-docker.sh` — core build + NSIS
-  packaging run inside the container (`USE_SYSTEM_WINE=true` uses the image's
-  wine); electron-builder's ~400 MB toolchain cache lives in a docker volume so
-  later builds reuse it.
-- The `modelrouter/win-builder` image is auto-built on first run (~5-10 min)
-  and reused afterwards.
-- After packaging, `packaging/verify-dist.sh` runs as a gate: an exe under
-  5 MB (or a deb/rpm/7z under 10 MB) fails the build — a broken installer can
-  never reach a Release again.
-
-```bash
-# Full Windows release (rebuild core + NSIS):
-make gui-win-docker
-```
-
-### Features
-
-- **Embedded core**: auto-spawns the bundled `llmsproxy` (LuaJIT build);
-  config/keys/adapters live in `<userData>/profile/`. First run generates a
-  random admin key written into `keys` (not seed) and self-injects it — WebUI
-  needs no login and never nags about changing the initial key. Port changeable
-  in settings (default 8787).
-- **System tray**: core status dot, autostart/silent toggles, start/stop/restart
-  core, quit.
-- **Silent launch**: `--silent` flag or setting — starts to the tray only;
-  first run always shows the window.
-- **Autostart**: `setLoginItemSettings` on Windows/macOS; writes
-  `~/.config/autostart/modelrouter-gui.desktop` on Linux, appending `--silent`
-  when silent mode is on.
+The plugin SDK follows the same naming: the only public entry point is
+`modelrouter` (`llmsproxy` remains a compatibility alias). See
+[docs/plugins.md](docs/plugins.md).
 
 ## Layout
 
-```
-cmd/llmsproxy            # entry point
-internal/config          # YAML config loading/validation
-internal/lua             # LuaJIT VM + worker pool + AdapterCache + built-ins (embed)
+```text
+cmd/modelrouter          # entry point (pure Go server)
+cmd/gui                  # Electron desktop shell (embeds the same core)
+internal/config          # config parsing, sealed storage, runtime file
+internal/core            # scheduling, kernel actions, state and audit
+internal/lua             # LuaJIT VM + elastic worker pool + built-in adapters (embed)
+internal/lua/plugins     # optional plugin examples (not shipped in binary/archives)
 internal/provider        # Provider(HTTP) + Registry(routing)
-internal/gateway         # OpenAI-compatible HTTP + auth + SDK/streaming + WebUI
-internal/types           # unified format & OpenAI wire types
+internal/gateway         # OpenAI-compatible HTTP service + auth + streaming + WebUI
+internal/types           # unified format and OpenAI wire types
+packaging/               # deb/rpm/tar packaging, systemd unit, migration script
+docs/                    # topic documentation
 ```
 
-## Development & collaboration
-
-- **Git branching workflow**: main / feature / release-branch conventions in
-  **[docs/git-workflow-en.md](docs/git-workflow-en.md)**
-  ([中文](docs/git-workflow.md)). The line: `main` is always deployable →
-  features merge into `main` → cut `release/vX.Y.Z` and tag → release → hotfixes
-  land on the release branch and are cherry-picked back into `main`.
-- **Lua adapter protocol**: see
-  **[docs/lua-adapters-en.md](docs/lua-adapters-en.md)**
-  ([中文](docs/lua-adapters.md)).
-
-## Tests
+## Development and tests
 
 ```bash
-go test -tags luajit ./...
+go test -tags luajit ./...           # everything
+go test -tags luajit ./internal/lua  # one package (required for LuaJIT changes)
 ```
 
-Covers: config validation, adapter load/transform, signature hooks, gateway
-auth, SDK round trip, SSE streaming, model routing, multimodal passthrough and
-disable_thinking.
+- Branch and release discipline: [docs/git-workflow-en.md](docs/git-workflow-en.md).
+- Plugins live in `plugin_dir`. The gateway never creates, overwrites or deletes
+  plugin files, and a desktop upgrade never touches them either.

cmd/gui/main.js

@@ -1,5 +1,5 @@
 // ModelRouter Desktop GUI — main process
-// Clash-Verge-style: bundles and manages an embedded llmsproxy core instance
+// Desktop shell: bundles and manages an embedded modelrouter core instance
 // in a per-user profile dir (userData), auto-starts it, keeps a tray icon,
 // supports launch-at-login. Server-side users keep using the pure Go binary.
 const {
@@ -50,11 +50,27 @@ const CONFIG_FILE = path.join(PROFILE_DIR, "config.yaml");
 const PLUGINS_DIR = path.join(PROFILE_DIR, "plugins");
 // Packaged: extraResource at process.resourcesPath/bin (outside asar, spawnable).
 // Dev:   cmd/gui/bin, populated by prepare-core.js.
-const CORE_EXE = path.join(
+// The embedded core is named after the product. Installs produced before the
+// llmsproxy -> modelrouter rename still carry the old file name inside their
+// profile, so resolution falls back to it instead of declaring the core missing.
+const CORE_BIN_DIR = path.join(
   app.isPackaged ? process.resourcesPath : __dirname,
   "bin",
+);
+const CORE_EXE = path.join(
+  CORE_BIN_DIR,
+  process.platform === "win32" ? "modelrouter.exe" : "modelrouter",
+);
+const LEGACY_CORE_EXE = path.join(
+  CORE_BIN_DIR,
   process.platform === "win32" ? "llmsproxy.exe" : "llmsproxy",
 );
+
+function resolveCoreExe() {
+  if (fs.existsSync(CORE_EXE)) return CORE_EXE;
+  if (fs.existsSync(LEGACY_CORE_EXE)) return LEGACY_CORE_EXE;
+  return CORE_EXE;
+}
 // Packaged window icon: must be a REAL file on disk (extraResource), not an
 // asar path — Electron/GTK pulls BrowserWindow icons from the filesystem and
 // crashes (SIGSEGV) when given an asar virtual path on Linux.
@@ -163,7 +179,7 @@ function unsealViaCore() {
   let key = "";
   try {
     const out = execFileSync(
-      CORE_EXE,
+      resolveCoreExe(),
       ["-config", CONFIG_FILE, "-show-secrets"],
       {
         encoding: "utf-8",
@@ -215,7 +231,7 @@ function coreStarted() {
 function writeDefaultEmbeddedConfig() {
   // Generate a self-contained embedded profile with a fresh random admin key.
   // No upstream sources are preconfigured — the desktop user adds them via
-  // the WebUI's sources page (like Clash Verge's empty-by-default profile).
+  // the WebUI's sources page (empty by default, like the WebUI profile).
   // If a config already exists we do NOT overwrite it (keep user's edits).
   if (fs.existsSync(CONFIG_FILE)) return;
   const key = "sk-gw-" + crypto.randomBytes(16).toString("hex");
@@ -272,8 +288,9 @@ function startCore() {
     console.log("core already running");
     return;
   }
-  if (!fs.existsSync(CORE_EXE)) {
-    console.error("embedded core binary missing:", CORE_EXE);
+  const coreExe = resolveCoreExe();
+  if (!fs.existsSync(coreExe)) {
+    console.error("embedded core binary missing:", coreExe);
     return;
   }
   // ensure profile dir + config + local adapters exist
@@ -281,9 +298,8 @@ function startCore() {
   if (!fs.existsSync(CONFIG_FILE)) writeDefaultEmbeddedConfig();
   pluginSync.ensurePluginDirConfigured(CONFIG_FILE, PLUGINS_DIR, log);
   fs.mkdirSync(PLUGINS_DIR, { recursive: true });
-  pluginSync.syncBundledPluginsOnUpgrade(PROFILE_DIR, PLUGINS_DIR, app.getVersion(), log);
 
-  coreProc = spawn(CORE_EXE, ["-config", CONFIG_FILE], {
+  coreProc = spawn(coreExe, ["-config", CONFIG_FILE], {
     cwd: PROFILE_DIR,
     env: Object.assign({}, process.env, {
       LLMS_PROXY_GUI: "1",
@@ -686,7 +702,7 @@ function createWindow() {
     });
   }
   // Silent start: `--silent` CLI flag or settings.silentStart keeps the app
-  // in the tray without showing the window (Clash-Verge style). First run
+  // in the tray without showing the window (desktop-shell behaviour). First run
   // always shows the window so the user sees the dashboard.
   mainWindow.once("ready-to-show", () => {
     const firstRun = !fs.existsSync(
@@ -810,8 +826,8 @@ ipcMain.handle("core:state", () => ({
   port: settings.port || DEFAULT_PORT,
   profileDir: PROFILE_DIR,
   configFile: CONFIG_FILE,
-  coreExe: CORE_EXE,
-  coreExists: fs.existsSync(CORE_EXE),
+  coreExe: resolveCoreExe(),
+  coreExists: fs.existsSync(resolveCoreExe()),
 }));
 ipcMain.handle("core:start", () => {
   app.isQuitting = false;
@@ -883,7 +899,7 @@ if (gotLock) {
     createWindow();
     createTray();
 
-    // start embedded core if not already running (another instance of llmsproxy)
+    // start embedded core if not already running (another instance of modelrouter)
     setTimeout(async () => {
       const p = await probeEmbedded();
       if (!p.ok) {

cmd/gui/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "modelrouter-gui",
-  "version": "2.0.44",
+  "version": "3.0.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "modelrouter-gui",
-      "version": "2.0.44",
+      "version": "3.0.0",
       "license": "Proprietary",
       "devDependencies": {
         "electron": "^33.0.0",

cmd/gui/package.json

@@ -1,6 +1,6 @@
 {
   "name": "modelrouter-gui",
-  "version": "2.0.44",
+  "version": "3.0.0",
   "description": "ModelRouter Desktop — embedded ModelRouter core with tray",
   "author": "ModelRouter",
   "main": "main.js",
@@ -39,8 +39,8 @@
       },
       "extraResources": [
         {
-          "from": "bin/llmsproxy",
-          "to": "bin/llmsproxy"
+          "from": "bin/modelrouter",
+          "to": "bin/modelrouter"
         },
         {
           "from": "build/icon.png",
@@ -66,8 +66,8 @@
       "icon": "build/icon.ico",
       "extraResources": [
         {
-          "from": "bin/llmsproxy.exe",
-          "to": "bin/llmsproxy.exe"
+          "from": "bin/modelrouter.exe",
+          "to": "bin/modelrouter.exe"
         },
         {
           "from": "bin/lua51.dll",

cmd/gui/plugin-sync.js

@@ -1,18 +1,11 @@
-// Plugin-tree maintenance for the embedded core (no Electron imports, so this
+// Plugin-tree maintenance for the desktop profile (no Electron imports, so this
 // is unit-testable with plain node).
 //
-// Two problems this solves, both invisible from the gateway side:
-//   1. The core deliberately does NOT default `plugin_dir` (an operator opts
-//      in). A desktop profile without it runs with ZERO plugins — the channels
-//      plugin (account login, credential vault, auto check-in) never appears.
-//   2. The core seeds bundled plugins only when the file is ABSENT, so a
-//      desktop upgrade would keep the user's old channels.lua forever and
-//      silently skip every channel fix. On version change the shell must drop
-//      the built-in files so the new binary reseeds them.
+// The desktop may configure an operator-selected plugin directory, but it never
+// owns the files in that directory. In particular, upgrades must not seed,
+// replace, rename, or delete plugins. Plugin installation is an explicit
+// operator action (or an external package manager action).
 const fs = require("fs");
-const path = require("path");
-
-const BUILTIN_PLUGINS = ["billing.lua", "channels.lua", "quota.lua", "usage.lua"];
 
 // ensurePluginDirConfigured back-fills `plugin_dir` into an existing config.
 // Appending a top-level YAML key is safe for the shapes this file can have
@@ -25,7 +18,7 @@ function ensurePluginDirConfigured(configFile, pluginsDir, log) {
     fs.writeFileSync(
       configFile,
       cur.replace(/\s*$/, "") +
-        "\n# added by ModelRouter Desktop: enables the plugin tree (channels etc.)\n" +
+        "\n# added by ModelRouter Desktop: enables the operator plugin directory\n" +
         "plugin_dir: " + JSON.stringify(pluginsDir) + "\n",
       "utf-8",
     );
@@ -37,46 +30,4 @@ function ensurePluginDirConfigured(configFile, pluginsDir, log) {
   }
 }
 
-// syncBundledPluginsOnUpgrade drops built-in plugin files when the app version
-// changed, so the core reseeds them from the binary it shipped with. Returns
-// { changed, dropped:[...], kept: "prev-<ver>" }.
-function syncBundledPluginsOnUpgrade(profileDir, pluginsDir, version, log) {
-  const stamp = path.join(profileDir, ".plugin-seed-version");
-  let have = "";
-  try {
-    have = fs.readFileSync(stamp, "utf-8").trim();
-  } catch (e) {
-    /* first run */
-  }
-  if (have === version) return { changed: false, dropped: [] };
-  const dropped = [];
-  for (const name of BUILTIN_PLUGINS) {
-    const f = path.join(pluginsDir, name);
-    if (!fs.existsSync(f)) continue;
-    try {
-      if (have) fs.renameSync(f, f + ".prev-" + have);
-      else fs.unlinkSync(f);
-      dropped.push(name);
-    } catch (e) {
-      if (log) log("[gui] drop builtin plugin " + name + " failed: " + e.message);
-    }
-  }
-  try {
-    fs.writeFileSync(stamp, version + "\n", "utf-8");
-  } catch (e) {
-    if (log) log("[gui] write plugin seed stamp failed: " + e.message);
-  }
-  if (dropped.length && log) {
-    log(
-      "[gui] plugins refreshed for " + version + " (" + dropped.length +
-        " builtin file(s) reseeded by core)",
-    );
-  }
-  return { changed: true, dropped, kept: have ? "prev-" + have : "" };
-}
-
-module.exports = {
-  BUILTIN_PLUGINS,
-  ensurePluginDirConfigured,
-  syncBundledPluginsOnUpgrade,
-};
+module.exports = { ensurePluginDirConfigured };

cmd/gui/plugin-sync.test.js

@@ -1,26 +1,22 @@
-// Unit tests for the desktop plugin-tree maintenance (plain node, no Electron).
-// Run: node --test cmd/gui/plugin-sync.test.js
+// Unit tests for desktop plugin-directory configuration (plain node).
 const test = require("node:test");
 const assert = require("node:assert");
 const fs = require("fs");
 const os = require("os");
 const path = require("path");
-const { BUILTIN_PLUGINS, ensurePluginDirConfigured, syncBundledPluginsOnUpgrade } = require("./plugin-sync");
+const { ensurePluginDirConfigured } = require("./plugin-sync");
 
 const tmp = () => fs.mkdtempSync(path.join(os.tmpdir(), "plugin-sync-"));
 
 test("back-fills plugin_dir into an existing profile without it", () => {
   const dir = tmp();
   const cfg = path.join(dir, "config.yaml");
-  fs.writeFileSync(cfg, 'listen: 127.0.0.1:8787\ndefault_model: AUTO\nkeys:\n  - key: k\n    role: admin\n');
+  fs.writeFileSync(cfg, "listen: 127.0.0.1:8787\ndefault_model: AUTO\nkeys:\n  - key: k\n    role: admin\n");
   const plugins = path.join(dir, "plugins");
   assert.equal(ensurePluginDirConfigured(cfg, plugins, null), "added");
   const out = fs.readFileSync(cfg, "utf-8");
   assert.match(out, /^plugin_dir: "/m);
-  assert.ok(out.includes(plugins.replace(/\\/g, "\\\\")) || out.includes(plugins), "path written");
-  // the rest of the profile must survive untouched
-  assert.match(out, /^listen: 127\.0\.0\.1:8787$/m);
-  assert.match(out, /^  - key: k$/m);
+  assert.match(out, /^listen: 127.0.0.1:8787$/m);
 });
 
 test("leaves a profile that already sets plugin_dir alone", () => {
@@ -39,42 +35,14 @@ test("missing config is not created by the back-fill", () => {
   assert.equal(fs.existsSync(cfg), false);
 });
 
-test("upgrade drops builtin plugins (keeping a backup) but not user plugins", () => {
+test("desktop upgrade never modifies operator plugin files", () => {
   const dir = tmp();
   const plugins = path.join(dir, "plugins");
   fs.mkdirSync(plugins, { recursive: true });
-  for (const f of BUILTIN_PLUGINS) fs.writeFileSync(path.join(plugins, f), "OLD");
+  fs.writeFileSync(path.join(plugins, "channels.lua"), "OPERATOR FILE");
   fs.writeFileSync(path.join(plugins, "my-own.lua"), "MINE");
-  fs.writeFileSync(path.join(dir, ".plugin-seed-version"), "1.0.0\n");
-
-  const r = syncBundledPluginsOnUpgrade(dir, plugins, "2.0.42", null);
-  assert.equal(r.changed, true);
-  assert.deepEqual(r.dropped.sort(), [...BUILTIN_PLUGINS].sort());
-  for (const f of BUILTIN_PLUGINS) {
-    assert.equal(fs.existsSync(path.join(plugins, f)), false, f + " dropped");
-    assert.equal(fs.readFileSync(path.join(plugins, f + ".prev-1.0.0"), "utf-8"), "OLD");
-  }
+  // There is deliberately no version stamp and no upgrade synchronizer.
+  assert.equal(fs.readFileSync(path.join(plugins, "channels.lua"), "utf-8"), "OPERATOR FILE");
   assert.equal(fs.readFileSync(path.join(plugins, "my-own.lua"), "utf-8"), "MINE");
-  assert.equal(fs.readFileSync(path.join(dir, ".plugin-seed-version"), "utf-8").trim(), "2.0.42");
-});
-
-test("same version is a no-op (nothing dropped)", () => {
-  const dir = tmp();
-  const plugins = path.join(dir, "plugins");
-  fs.mkdirSync(plugins, { recursive: true });
-  fs.writeFileSync(path.join(plugins, "channels.lua"), "CURRENT");
-  fs.writeFileSync(path.join(dir, ".plugin-seed-version"), "2.0.42\n");
-  const r = syncBundledPluginsOnUpgrade(dir, plugins, "2.0.42", null);
-  assert.equal(r.changed, false);
-  assert.equal(fs.readFileSync(path.join(plugins, "channels.lua"), "utf-8"), "CURRENT");
-});
-
-test("first run on a fresh profile just stamps the version", () => {
-  const dir = tmp();
-  const plugins = path.join(dir, "plugins");
-  fs.mkdirSync(plugins, { recursive: true });
-  const r = syncBundledPluginsOnUpgrade(dir, plugins, "2.0.42", null);
-  assert.equal(r.changed, true);
-  assert.deepEqual(r.dropped, []);
-  assert.equal(fs.existsSync(path.join(dir, ".plugin-seed-version")), true);
+  assert.equal(fs.readdirSync(dir).includes(".plugin-seed-version"), false);
 });

cmd/gui/renderer/app.js

@@ -1,5 +1,5 @@
 // ModelRouter Desktop GUI — renderer
-// Clash-Verge-style: the content area is a full-screen iframe loading the
+// Desktop shell: the content area is a full-screen iframe loading the
 // real WebUI 1:1 (status / chat / keys / priority / sources / adapters).
 // The shell only adds the titlebar, settings overlay and tray integration.
 const state = {

cmd/gui/renderer/index.html

@@ -83,7 +83,7 @@
       </div>
     </div>
 
-    <!-- side rail: autostart / silent toggles (Clash-Verge style) -->
+    <!-- side rail: autostart / silent toggles (desktop-shell style) -->
     <aside class="rail" id="rail">
       <div class="rail-gap"></div>
       <button

cmd/gui/renderer/style.css

@@ -1,5 +1,5 @@
 /* =====================================================================
-   ModelRouter Desktop GUI — Clash-Verge-style shell
+   ModelRouter Desktop GUI — desktop-shell-style shell
    Content area = full-screen iframe embedding the real WebUI 1:1.
    Palette mirrors the WebUI (sakura/frost/glassmorphism).
    ===================================================================== */

cmd/gui/scripts/prepare-core.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-// Build the embedded llmsproxy core(s) for the GUI package.
+// Build the embedded modelrouter core(s) for the GUI package.
 //   node prepare-core.js        -> linux (native, luajit) + win (cross, best effort)
 //   node prepare-core.js linux  -> linux only
 //   node prepare-core.js win    -> win only
@@ -12,21 +12,21 @@ const binDir = path.join(guiDir, "bin");
 const projectRoot = path.join(guiDir, "..", "..");
 
 function buildLinux() {
-  const out = path.join(binDir, "llmsproxy");
-  console.log("[prepare-core] building llmsproxy (linux/luajit)...");
+  const out = path.join(binDir, "modelrouter");
+  console.log("[prepare-core] building modelrouter (linux/luajit)...");
   fs.mkdirSync(binDir, { recursive: true });
   execFileSync(
     "go",
-    ["build", "-tags", "luajit", "-trimpath", "-o", out, "./cmd/llmsproxy"],
+    ["build", "-tags", "luajit", "-trimpath", "-o", out, "./cmd/modelrouter"],
     { cwd: projectRoot, stdio: "inherit" },
   );
   return out;
 }
 
 function buildWin() {
-  const out = path.join(binDir, "llmsproxy.exe");
+  const out = path.join(binDir, "modelrouter.exe");
   console.log(
-    "[prepare-core] cross-building llmsproxy.exe (windows/amd64, luajit)...",
+    "[prepare-core] cross-building modelrouter.exe (windows/amd64, luajit)...",
   );
   fs.mkdirSync(binDir, { recursive: true });
   const cc = process.env.MINGW_CC || "x86_64-w64-mingw32-gcc";
@@ -43,7 +43,7 @@ function buildWin() {
   }
   execFileSync(
     "go",
-    ["build", "-tags", "luajit", "-trimpath", "-o", out, "./cmd/llmsproxy"],
+    ["build", "-tags", "luajit", "-trimpath", "-o", out, "./cmd/modelrouter"],
     {
       cwd: projectRoot,
       stdio: "inherit",

cmd/modelrouter/main.go

@@ -1,4 +1,4 @@
-// Command llmsproxy is a standalone gateway that exposes multiple upstream LLM
+// Command modelrouter is a standalone gateway that exposes multiple upstream LLM
 // sources behind a unified OpenAI-compatible HTTP API. Protocol differences are
 // handled by Lua adapters (per source), optionally signing outgoing requests.
 // It supports explicit model selection or AUTO routing by priority, image
@@ -16,9 +16,9 @@ import (
 	"syscall"
 	"time"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/core"
-	"llmsproxy/internal/gateway"
+	"modelrouter/internal/config"
+	"modelrouter/internal/core"
+	"modelrouter/internal/gateway"
 )
 
 func main() {
@@ -33,12 +33,12 @@ func main() {
 	// error here instead of being silently created.
 	if *checkOnly {
 		if _, err := os.Stat(*cfgPath); err != nil {
-			log.Fatalf("[llmsproxy] check: %v", err)
+			log.Fatalf("[modelrouter] check: %v", err)
 		}
 		if _, err := config.Load(*cfgPath); err != nil {
-			log.Fatalf("[llmsproxy] check: %v", err)
+			log.Fatalf("[modelrouter] check: %v", err)
 		}
-		log.Printf("[llmsproxy] check: %s is valid", *cfgPath)
+		log.Printf("[modelrouter] check: %s is valid", *cfgPath)
 		return
 	}
 
@@ -48,34 +48,34 @@ func main() {
 	// program prints a credential, and this path never logs to a file.
 	if *showSecrets {
 		if _, err := os.Stat(*cfgPath); err != nil {
-			log.Fatalf("[llmsproxy] show-secrets: %v", err)
+			log.Fatalf("[modelrouter] show-secrets: %v", err)
 		}
 		if err := config.PrintSecrets(*cfgPath); err != nil {
-			log.Fatalf("[llmsproxy] show-secrets: %v", err)
+			log.Fatalf("[modelrouter] show-secrets: %v", err)
 		}
 		return
 	}
 
 	created, err := config.EnsureDefault(*cfgPath)
 	if err != nil {
-		log.Fatalf("[llmsproxy] config: %v", err)
+		log.Fatalf("[modelrouter] config: %v", err)
 	}
 
 	c, err := core.New(*cfgPath)
 	if err != nil {
-		log.Fatalf("[llmsproxy] core: %v", err)
+		log.Fatalf("[modelrouter] core: %v", err)
 	}
 	defer c.Close()
 
 	if created {
 		for _, k := range c.GatewayKeys() {
-			log.Printf("[llmsproxy] generated default config at %s — admin key: %s (rotate it in the WebUI after first login)", *cfgPath, k)
+			log.Printf("[modelrouter] generated default config at %s — admin key: %s (rotate it in the WebUI after first login)", *cfgPath, k)
 		}
 	}
 
 	gw, err := gateway.New(c)
 	if err != nil {
-		log.Fatalf("[llmsproxy] gateway: %v", err)
+		log.Fatalf("[modelrouter] gateway: %v", err)
 	}
 
 	// Ops hygiene: surface the two most common footguns instead of silently
@@ -88,16 +88,16 @@ func main() {
 	// (e.g. after rotating the starter key away) it reported "every request is
 	// rejected" while seven working keys, one of them admin, were in service.
 	if keys := c.ListKeys(); len(keys) == 0 {
-		log.Printf("[llmsproxy] WARNING: no gateway key configured — every request will be rejected. Add one in the WebUI or set gateway_keys")
+		log.Printf("[modelrouter] WARNING: no gateway key configured — every request will be rejected. Add one in the WebUI or set gateway_keys")
 	} else {
 		for _, k := range keys {
 			if k.Key == "sk-gw-local-0001" || k.Key == "sk-local-0001" {
-				log.Printf("[llmsproxy] WARNING: gateway key %q looks like the starter/example key — rotate it before exposing the gateway", k.Key)
+				log.Printf("[modelrouter] WARNING: gateway key %q looks like the starter/example key — rotate it before exposing the gateway", k.Key)
 			}
 		}
 	}
 	if l := c.Listen(); strings.HasPrefix(l, "0.0.0.0:") || strings.HasPrefix(l, "::") {
-		log.Printf("[llmsproxy] WARNING: listen=%s binds ALL interfaces — bind an internal address in production", l)
+		log.Printf("[modelrouter] WARNING: listen=%s binds ALL interfaces — bind an internal address in production", l)
 	}
 
 	srv := &http.Server{
@@ -108,22 +108,22 @@ func main() {
 	go func() {
 		cert, key := c.TLS()
 		if cert != "" && key != "" {
-			log.Printf("[llmsproxy] listening HTTPS on %s (cert=%q) (default_model=%s, models=%v, adapters=%d)",
+			log.Printf("[modelrouter] listening HTTPS on %s (cert=%q) (default_model=%s, models=%v, adapters=%d)",
 				c.Listen(), cert, c.DefaultModel(), c.Registry().ModelList(), len(c.ListAdapters()))
 			if err := srv.ListenAndServeTLS(cert, key); err != nil && err != http.ErrServerClosed {
-				log.Fatalf("[llmsproxy] server: %v", err)
+				log.Fatalf("[modelrouter] server: %v", err)
 			}
 			return
 		}
-		log.Printf("[llmsproxy] listening HTTP on %s (default_model=%s, models=%v, adapters=%d)",
+		log.Printf("[modelrouter] listening HTTP on %s (default_model=%s, models=%v, adapters=%d)",
 			c.Listen(), c.DefaultModel(), c.Registry().ModelList(), len(c.ListAdapters()))
 		if err := srv.ListenAndServe(); err != nil && err != http.ErrServerClosed {
-			log.Fatalf("[llmsproxy] server: %v", err)
+			log.Fatalf("[modelrouter] server: %v", err)
 		}
 	}()
 
 	stop := make(chan os.Signal, 1)
 	signal.Notify(stop, syscall.SIGINT, syscall.SIGTERM)
 	<-stop
-	log.Printf("[llmsproxy] shutting down")
+	log.Printf("[modelrouter] shutting down")
 }

deploy.sh

@@ -1,10 +1,10 @@
 #!/usr/bin/env bash
 # ============================================================
-# llmsproxy 原子化滚动部署脚本
+# modelrouter 原子化滚动部署脚本
 #
 # 策略：
 #   1. /tmp (tmpfs) 构建，/usr/local/bin (ext4) 部署
-#   2. 构建后先 copy 到目标文件系统（.llmsproxy.tmp），
+#   2. 构建后先 copy 到目标文件系统（.modelrouter.tmp），
 #      再 rename(2) 做原子替换（同一文件系统内 mv 是原子的）
 #   3. runtime.json 不替换（保留运行时密钥、审计状态）
 #   4. systemctl restart → 服务中断仅在 Go 进程重启瞬间 (<500ms)
@@ -28,20 +28,21 @@
 set -euo pipefail
 
 # ---------- 配置 ----------
-readonly TARGET_BIN="/usr/local/bin/llmsproxy"
-readonly TARGET_CONFIG="/etc/llmsproxy/config.yaml"
-readonly TARGET_ADAPTERS="/etc/llmsproxy/adapters"
-readonly TARGET_PLUGINS="/etc/llmsproxy/plugins"
-readonly REPO_DIR="/home/program/llmsproxy"
+readonly TARGET_BIN="/usr/local/bin/modelrouter"
+readonly TARGET_CONFIG="/etc/modelrouter/config.yaml"
+readonly TARGET_ADAPTERS="/etc/modelrouter/adapters"
+readonly TARGET_PLUGINS="/etc/modelrouter/plugins"
+# 仓库所在目录不等于产品名：这里保持真实路径，除非用 MODELROUTER_REPO_DIR 覆盖。
+readonly REPO_DIR="${MODELROUTER_REPO_DIR:-/home/program/llmsproxy}"
 readonly HEALTH_URL="http://127.0.0.1:8081/v1/models"
-readonly BIN_NAME="llmsproxy"
+readonly BIN_NAME="modelrouter"
 
 # 构建目录。默认落在仓库下的构建工作区，而不是 /tmp：本机 /tmp 是 8.8G
 # tmpfs，实测被 Go 构建缓存占满后会让 go test/build 在链接阶段直接失败。
-# 可用 LLMSPROXY_BUILD_DIR 覆盖；同一目录也被测试用作 GOTMPDIR。
+# 可用 MODELROUTER_BUILD_DIR 覆盖；同一目录也被测试用作 GOTMPDIR。
 readonly BUILD_SUBDIR=".build-work"
-if [[ -n "${LLMSPROXY_BUILD_DIR:-}" ]]; then
-    readonly TMP_DIR="${LLMSPROXY_BUILD_DIR}"
+if [[ -n "${MODELROUTER_BUILD_DIR:-}" ]]; then
+    readonly TMP_DIR="${MODELROUTER_BUILD_DIR}"
 else
     readonly TMP_DIR="${REPO_DIR}/${BUILD_SUBDIR}"
 fi
@@ -59,7 +60,7 @@ usage() {
     cat <<'EOF'
 用法: ./deploy.sh [--config <file>]
 
-  --config <file>   投放这份配置到 /etc/llmsproxy/config.yaml（原子替换）。
+  --config <file>   投放这份配置到 /etc/modelrouter/config.yaml（原子替换）。
                     省略时不改动线上配置，只部署二进制与适配器。
                     无论是否指定，重启前都会用新二进制 -check 预检线上配置。
 
@@ -103,6 +104,17 @@ precheck() {
     command -v systemctl &>/dev/null || fail "未找到 systemctl"
     [[ $(id -u) -eq 0 ]] || fail "需要 root 权限"
 
+    # 改名保护：本脚本现在部署到 /etc/modelrouter。若机器上仍是改名前的
+    # /etc/llmsproxy 且新路径尚未建立，直接继续会创建一份**空配置**并让新
+    # 服务在没有 key/源的情况下起来 —— 看起来成功，实际网关拒掉所有请求。
+    if [[ -d /etc/llmsproxy && ! -e "$TARGET_CONFIG" ]]; then
+        fail "检测到改名前的部署 /etc/llmsproxy，但 $TARGET_CONFIG 不存在。
+     先迁移（dry-run 默认，确认后加 --apply）：
+       sudo ./packaging/migrate-to-modelrouter.sh
+       sudo ./packaging/migrate-to-modelrouter.sh --apply
+     如需全新部署，先把旧目录移开：mv /etc/llmsproxy /etc/llmsproxy.bak"
+    fi
+
     if [[ -n "$NEW_CONFIG_SRC" ]]; then
         [[ -f "$NEW_CONFIG_SRC" ]] || fail "待部署配置不存在: $NEW_CONFIG_SRC"
         # 同一文件比对：--config 指向线上文件本身时没有"投放"动作可做，
@@ -153,7 +165,7 @@ build() {
         -trimpath \
         -ldflags="-s -w" \
         -o "$TMP_DIR/bin/$BIN_NAME" \
-        ./cmd/llmsproxy
+        ./cmd/modelrouter
 
     NEW_BIN="$TMP_DIR/bin/$BIN_NAME"
     NEW_SIZE=$(stat -c%s "$NEW_BIN")
@@ -254,7 +266,7 @@ verify_config() {
 
 # ---------- 密钥可解封校验 ----------
 # config.yaml 里的凭据以 enc:v1: 密文落盘，解密钥匙是 master.key（或
-# LLMS_PROXY_MASTER_KEY）。一旦钥匙丢失/错位，配置在**启动时**才会报错，
+# MODELROUTER_MASTER_KEY）。一旦钥匙丢失/错位，配置在**启动时**才会报错，
 # 那时服务已经处于「二进制已替换 + 配置已替换」的半死状态，只能靠回滚。
 # 所以在替换任何东西之前先验：能不能用现有的钥匙解开现有的密文。
 #
@@ -279,21 +291,22 @@ verify_master_key() {
     [[ -n "$runtime_file" ]] || runtime_file="$(dirname "$TARGET_CONFIG")/runtime.json"
     key_file="$(dirname "$runtime_file")/master.key"
 
-    if [[ -n "${LLMS_PROXY_MASTER_KEY:-}" ]]; then
-        log "  钥匙来源: LLMS_PROXY_MASTER_KEY（环境变量）"
+    MASTER_KEY_ENV="${MODELROUTER_MASTER_KEY:-${LLMS_PROXY_MASTER_KEY:-}}"
+    if [[ -n "$MASTER_KEY_ENV" ]]; then
+        log "  钥匙来源: MODELROUTER_MASTER_KEY（环境变量）"
         # 走真实解密路径验证，而不是只检查格式
         if out=$("$TARGET_BIN" -show-secrets -config "$TARGET_CONFIG" 2>&1); then
             log "OK 凭据可解封（$(echo "$out" | grep -cE '^(source|key) ') 项）"
             return 0
         fi
-        err "凭据无法用 LLMS_PROXY_MASTER_KEY 解封，未替换任何文件："
+        err "凭据无法用 MODELROUTER_MASTER_KEY 解封，未替换任何文件："
         printf '%s\n' "$out" >&2
         return 1
     fi
 
     if [[ ! -f "$key_file" ]]; then
         err "配置含 $sealed 处密文，但钥匙文件不存在: $key_file"
-        err "  没有它，部署后服务将无法启动。恢复 master.key 或设置 LLMS_PROXY_MASTER_KEY 后重试。"
+        err "  没有它，部署后服务将无法启动。恢复 master.key 或设置 MODELROUTER_MASTER_KEY 后重试。"
         return 1
     fi
     log "  钥匙文件: $key_file ($(stat -c '%A' "$key_file" 2>/dev/null || echo '?')，$(wc -c < "$key_file") 字节)"
@@ -312,7 +325,7 @@ verify_master_key() {
 # ---------- 备份目录保留策略 ----------
 # 每次部署都新建一个 <prefix>.bak.<时间戳> 目录，而回滚只用到最近一次
 # （$BACKUP_BIN / $BACKUP_CONFIG 都是单文件覆盖）。多出来的目录从没人清理，
-# 实测线上已积累 71 个，/etc/llmsproxy 因此涨到 122M。
+# 实测线上已积累 71 个，/etc/modelrouter 因此涨到 122M。
 #
 # 保留最近 KEEP_BACKUPS 份足够回滚，同时给目录数设上限——否则一次误配置
 # 就可能在几秒内造出成百上千个目录。
@@ -327,7 +340,7 @@ KEEP_BACKUPS=5
 
 prune_backups() {
     local prefix="$1"
-    local base="/etc/llmsproxy"
+    local base="/etc/modelrouter"
     [[ -d "$base" ]] || return 0
     # 防误用：前缀必须是已知的两个（避免有人传空串删掉整个目录）
     case "$prefix" in
@@ -393,7 +406,7 @@ sync_adapters() {
     [[ -d "$SRC_ADAPTERS" ]] || fail "内嵌适配器目录不存在: $SRC_ADAPTERS"
 
     # 备份旧的适配器文件（含任何运行时覆盖版本，如 opencode.lua）
-    BACKUP_DIR="/etc/llmsproxy/adapters.bak.$(date +%Y%m%d%H%M%S)"
+    BACKUP_DIR="/etc/modelrouter/adapters.bak.$(date +%Y%m%d%H%M%S)"
     if [[ -d "$TARGET_ADAPTERS" ]]; then
         mkdir -p "$BACKUP_DIR"
         cp -rf "$TARGET_ADAPTERS/"*.lua "$BACKUP_DIR/" 2>/dev/null || true
@@ -425,7 +438,7 @@ sync_adapters() {
 #
 # 早期实现是「只补缺，已存在一律不动」，看着安全，实际造成一个功能性缺口：
 # 内置插件本身的**修复无法下发**。v2.0.3/v2.0.4/v2.0.5 三个渠道修复就这样
-# 全部卡在仓库里 —— 二进制换了、/etc/llmsproxy/plugins/channels.lua 还是旧版，
+# 全部卡在仓库里 —— 二进制换了、/etc/modelrouter/plugins/channels.lua 还是旧版，
 # 生产行为毫无变化。内置源码的正确性由仓库保证，部署必须与二进制内容对齐。
 #
 # 「只补缺」原本想保护的是「运维改过 / 运行时新增的插件」，但那些都是**非内置
@@ -439,7 +452,7 @@ sync_plugins() {
     mkdir -p "$TARGET_PLUGINS"
 
     # 备份将被覆盖/新增的内置插件（非内置名字不进入循环，天然不动）
-    BACKUP_DIR="/etc/llmsproxy/plugins.bak.$(date +%Y%m%d%H%M%S)"
+    BACKUP_DIR="/etc/modelrouter/plugins.bak.$(date +%Y%m%d%H%M%S)"
     mkdir -p "$BACKUP_DIR"
     local added=0 updated=0 same=0
     for src in "$SRC_PLUGINS"/*.lua; do
@@ -481,8 +494,8 @@ sync_plugins() {
 # ---------- 重启服务 ----------
 restart_service() {
     log "重启服务（中断 <500ms）"
-    systemctl restart llmsproxy.service || fail "systemd restart 失败"
-    log "OK llmsproxy.service restarted"
+    systemctl restart modelrouter.service || fail "systemd restart 失败"
+    log "OK modelrouter.service restarted"
 }
 
 # ---------- 健康检查 ----------
@@ -505,7 +518,7 @@ healthcheck() {
         (( attempt++ ))
     done
 
-    warn "健康检查未通过，请检查服务状态: systemctl status llmsproxy.service"
+    warn "健康检查未通过，请检查服务状态: systemctl status modelrouter.service"
     return 1
 }
 
@@ -541,7 +554,7 @@ rollback() {
 
     if (( restored_any == 0 )); then
         err "无任何回滚点，无法自动恢复"
-        err "请手动检查: systemctl status llmsproxy.service && journalctl -u llmsproxy -n 50"
+        err "请手动检查: systemctl status modelrouter.service && journalctl -u modelrouter -n 50"
         return 1
     fi
 
@@ -555,9 +568,9 @@ rollback() {
     fi
 
     log "重启服务（旧版本）"
-    systemctl restart llmsproxy.service || {
+    systemctl restart modelrouter.service || {
         err "重启失败，服务可能完全宕机"
-        err "请手动检查: systemctl status llmsproxy.service"
+        err "请手动检查: systemctl status modelrouter.service"
         return 1
     }
 
@@ -592,7 +605,7 @@ print_summary() {
 # 于是坏配置在服务仍然健康时就被拦下，根本不会触发重启。
 main() {
     parse_args "$@"
-    log "═══ llmsproxy 原子部署开始 ═══"
+    log "═══ modelrouter 原子部署开始 ═══"
     precheck
     # 密钥校验放在 build/替换之前：它的失败后果最重（配置里的密文永久解不开），
     # 必须在动任何文件之前就挡住。此阶段用的是**当前线上二进制**，它已经

docs/adapters.md

@@ -0,0 +1,60 @@
+# 适配器协议
+
+> 本文档由 README 拆分而来：Lua 适配器协议与内置适配器清单。
+> 编写适配器的完整参考见 [lua-adapters.md](lua-adapters.md)。
+> 返回 [README](../README.md)。
+
+## Lua 适配器协议
+
+完整 API 见 **[Lua 适配器 API 文档](docs/lua-adapters.md)**（[English](docs/lua-adapters-en.md)）。
+
+每个适配器是一个返回 table 的 Lua 脚本（`internal/lua/adapters/<name>.lua`），
+加载时可被 `adapter_dir` 下的同名脚本覆盖——**改 Lua 脚本无需重编译**，
+重启即生效；通过 WebUI 上传的适配器与在线编辑的源配置则即时生效。
+
+```lua
+return {
+  name = "mysrc",
+  version = "1.0.0",
+  endpoint = "/chat/completions",        -- 上游路径（可被 source.endpoint 覆盖）
+  headers = { ["X-Static"] = "v" },      -- 静态默认请求头（build_headers 缺省时使用）
+
+  -- 请求转换：把 OpenAI 格式 req 转成上游原生格式，返回字符串
+  transform_request = function(raw_json) ... end,
+
+  -- 响应转换：把上游原生响应转成统一格式字符串
+  --   { content, reasoning_content, finish_reason, token_usage{...}, tool_calls[{...}] }
+  transform_response = function(raw_json) ... end,
+
+  -- 流式分块转换：把上游 SSE data 转成 { content, done, ... }，返回 "" 则跳过
+  transform_stream_chunk = function(raw_chunk) ... end,
+
+  -- [可选] 动态请求头/签名钩子
+  -- meta = { url, method, body, api_key, timestamp, source={ name, meta={...} } }
+  build_headers = function(meta) return { ["X-App-Sign"] = sign } end,
+
+  -- [可选] 出向序列化（契约 v2）：上游不是 OpenAI 协议时接管写给客户端的字节。
+  -- 不声明则沿用内核的 OpenAI 兜底；声明即完全取代，内核不再混合字段。
+  serialize_response = function(result) return body_str end,
+  serialize_open_preamble = function(result) return bytes end,
+  serialize_chunk = function(delta) return bytes end,
+  serialize_close = function(result) return bytes end,
+}
+```
+
+内置信号辅助：`hmac_sha256_hex(key, data)`、`sha256_hex(data)`、`base64_encode(s)`、
+`tohex(s)`、`json.encode/decode`、`log(level, msg)`。
+
+### 内置适配器
+
+`openai` `deepseek` `anthropic` `gemini` `github` `groq` `mistral` `ollama` `kimicode`
+`qwen38` `sensenova` `trae`。
+
+`anthropic`/`gemini`/`ollama` 适配器内置多模态转换（`image_url` → 各自上游格式）；
+推理开关由中立的 `thinking` 字段表达，`deepseek` 适配器把它映射为
+`thinking.type = "disabled"`。
+
+**kimicode** 是展示 `build_headers` 的样例：云端校验调用方 app，需要按
+`meta.app_secret` 对时间戳+URL+请求体哈希做 HMAC 签名并附 `X-App-Sign` 等头。
+配好 `sources[].meta.{app_id, app_secret, app_agent}` 即可。
+

docs/channels-parity.md

@@ -389,10 +389,10 @@ flow 类型，不是"登录成功"。
 channels 插件用 `storeDir()` 决定存储位置，它有多级兜底：
 
 ```
-env LLMSPROXY_PLUGIN_DIR → settings.store_dir → 读 /etc/llmsproxy/config.yaml 的 plugin_dir → 硬编码
+env MODELROUTER_PLUGIN_DIR → settings.store_dir → 读 /etc/llmsproxy/config.yaml 的 plugin_dir → 硬编码
 ```
 
-测试台没设 `LLMSPROXY_PLUGIN_DIR`，**后两级直接命中生产**：
+测试台没设 `MODELROUTER_PLUGIN_DIR`，**后两级直接命中生产**：
 
 ```
 测试插件目录 = /tmp/TestXXX/002/plugins
@@ -404,7 +404,7 @@ storeDir 结果 = /etc/llmsproxy/plugins          ← 探针实测输出
 `.channels-store.json` 从 1730 覆盖成 1986 字节、账号池清空（inode 都换了）。
 
 只有 `internal/lua` 复现，`internal/core` 不复现 —— 因为 core 的测试台
-（`kernelTestCore`）设了 `LLMSPROXY_PLUGIN_DIR`。**这是隔离做得不一致**的代价。
+（`kernelTestCore`）设了 `MODELROUTER_PLUGIN_DIR`。**这是隔离做得不一致**的代价。
 
 同类前科：曾把 26 个测试账号写进生产 store。两次同一类错误。
 

docs/configuration.md

@@ -0,0 +1,183 @@
+# 配置
+
+> 本文档由 README 拆分而来：网关配置、模型自动探测、多密钥。
+> 返回 [README](../README.md)。
+
+## 配置文件
+
+首次运行在 `-config` 路径自动生成默认配置（默认 `./config.yaml`），核心字段：
+
+```yaml
+listen: 127.0.0.1:8080        # 网关监听地址（建议绑内网/回环）
+gateway_keys: [sk-gw-<随机>]  # 初始 admin 密钥（每次安装随机生成，仅首启写入运行时存储用）
+default_model: AUTO           # model 无法路由时自动按优先级链选源
+adapter_dir: adapters         # Lua 适配目录；目录不存在时首启创建并 seed 内置，存在则只读
+runtime_file: runtime.json    # WebUI 编辑的源/密钥/AUTO 链持久化到此文件
+
+sources:
+  - name: deepseek
+    base_url: https://api.deepseek.com
+    api_key: sk-...            # 也可用 api_key_env: SOME_ENV 引用环境变量（不落盘明文）
+    adapter: deepseek
+    max_concurrent: 8
+    models:
+      - id: deepseek-v4-flash
+        priority: 100          # YAML 源首次启动会 seed 进 AUTO 链的初值
+        kind: chat
+      - id: deepseek-v4-pro
+        priority: 60
+        kind: chat
+    # 静态请求头（比适配器默认优先）
+    headers: { X-Tenant: prod }
+    # 传给 Lua build_headers 的透传元数据
+    meta: { app_id: x, app_secret: y }
+    temperature: 0.7
+    max_tokens: 4096
+    timeout: 120s              # 请求超时，默认 120s
+```
+
+
+## 自动探测模型列表
+
+不想手抄模型 id 时，只填 URL 和 key，让网关去扫：
+
+```bash
+# 只探测，不写任何东西
+curl -X POST http://127.0.0.1:8080/api/sources/discover \
+  -H "Authorization: Bearer sk-gw-…" -H "Content-Type: application/json" \
+  -d '{"base_url":"https://api.vendor.com/v1","api_key":"sk-…"}'
+# → {"ok":true,"base_url":"…","endpoint":"","models":["gpt-x",…],
+#    "image_models":["gpt-image-2"],"models_path":"/models","latency_ms":1387,…}
+
+# 探测后直接落库
+curl -X POST http://127.0.0.1:8080/api/sources/discover/apply \
+  -H "Authorization: Bearer sk-gw-…" -H "Content-Type: application/json" \
+  -d '{"base_url":"https://api.vendor.com/v1","api_key":"sk-…","name":"vendor"}'
+```
+
+WebUI 源编辑框里也有「自动探测模型」按钮：填好 URL 和 key 点一下，列表自动
+填好，仍需点保存才生效。
+
+几点值得知道：
+
+- **URL 形态不用猜。** 厂商对「`/v1` 放 base 还是放 path」没有共识，探测会
+  依次尝试若干候选并返回答对的那个，同时给出归一化后的 `base_url` +
+  `endpoint`（`endpoint` 平时留空，因为 openai adapter 自带 `/chat/completions`）。
+- **`image_endpoint` 是另一套规则。** 网关的图像默认路径是**绝对**的
+  `/v1/images/generations`，而 chat 默认是相对的。所以 `base_url` 带 `/v1` 时
+  必须显式给 `image_endpoint: /images/generations`，否则图像请求 404。
+  探测会替你填好。
+- **图像模型只看上游自己声明的 modality**，绝不从模型名猜。`gpt-image-2`
+  这类 id 也出现在同时提供对话服务的目录上，猜错会让 AUTO 链选中一个只支持
+  图像生成的模型，其后每一层都被它拖垮。
+- **不会逐模型探活。** 34 个模型的目录跑一轮要花钱且耗时数分钟，`/models`
+  是足够便宜的信号；模型是否真能应答留给正常请求去发现。
+- **对已存在的源，必须显式 `replace_models: true`** 才会覆盖模型列表。重新
+  探测顺手抹掉为 AUTO 链手工调过的 priority，是最难排查的一类数据丢失。
+- **`api_key` 留空表示沿用已存值**，不会把凭据清空。
+
+
+`config.yaml`、`runtime.json`、`master.key`、`adapters/` 均不提交版本库（见
+`.gitignore`），请随备份带走——`master.key` 丢失后运行时密文无法解密。
+
+运行文件生成在 `-config` 所在目录：`adapter_dir` / `runtime_file` 自动指向
+配置文件旁边的 `adapters/` 与 `runtime.json`，从任意工作目录启动都可用。
+
+运行时文件（`runtime_file`）中的敏感字段自动加密：
+
+- 加密算法 AES-256-GCM，格式 `enc:v1:<base64>`。
+- 主密钥来源：环境变量 `MODELROUTER_MASTER_KEY`（64 位 hex）；否则读取
+  `runtime_file` 同目录的 `master.key`；都不存在时首次启动自动生成 `master.key`（0600）。
+- 升级时旧明文文件自动兼容：首次运行正常读取，任何 UI 保存操作触发全文件加密迁移。
+- 注意：**master.key 丢失后密文无法解密**，请随配置一起备份；切勿提交到版本库。
+
+## 网关密钥（多密钥）
+
+网关鉴权采用「多密钥 + 角色 + 模型范围」架构，密钥持久化在 `runtime_file` 的
+`keys` 字段（加密存储）：
+
+- `gateway_keys` 配置只是**初始 admin 密钥种子**：首次启动迁移为运行时 admin
+  key，之后不再参与鉴权管理。
+- **重要：首次启动后请在 WebUI「密钥」页更换管理员密钥**——初始密钥明文写在
+  `config.yaml` 里（每次安装随机生成），继续使用存在被盗风险；用新密钥登录后删除初始密钥。
+- WebUI **密钥页**可创建/删除密钥；每个密钥可指定 `admin`（管理全部）或
+  `user`（仅看自己的 key）角色，并配置**模型范围**（模型 + 源 + token 配额 +
+  重置周期）。
+- 客户端用任意一个已授权的密钥明文作为 Bearer（`Authorization: Bearer <key>`）。
+- 删除密钥即从运行时存储移除，立即失效。
+
+### 密钥用量配额
+
+**配额按模型单独设置**：每个密钥的 `models[]` 里，每一条模型范围各自带一份
+token 配额与请求数配额。一个模型用满只影响该模型，同一密钥的其它模型照常工作。
+
+```yaml
+keys:
+  - key: sk-gw-<hex>
+    role: user
+    name: agent-alice
+    models:
+      - model: deepseek-v4-flash
+        token_quota: 1000000   # 本周期内该 key 用这个模型的 token 预算
+        req_quota: 20000       # 本周期内的请求次数
+        period: nhour          # "" | hour | week | month | nhour
+        hours: 6               # 仅 nhour
+      - model: AUTO
+        token_quota: 5000000   # AUTO 也是一条独立配额
+        period: day            # ← 这种写法会被拒绝（词表只有 hour/week/month/nhour）
+      - model: kimi-k3        # 未列配额 = 无限
+```
+
+- **没有「整钥总配额」**：这是刻意的设计。整钥总额会让「把 A 模型的额度挪给
+  B 模型」变成一次全局重分配；按模型独立计费则每个模型各自可控，运维可以
+  看出哪个模型吃掉了预算。
+- `period` 词表：空 = 永不过期（累计总量），`hour` / `week` / `month` = 固定窗口，
+  `nhour` + `hours` = 自定义小时数。**拼错的周期在写入时就被拒**，不会静默变成
+  永不过期。
+- 配额严格按密钥隔离，且**同一密钥内按模型隔离**：A 密钥用满 `m1` 不会消耗
+  B 密钥的额度，同一密钥的 `m2` 也不受影响。
+- 配额统计含聊天、流式、生图，跨重启从审计日志回放（保留 40 天，覆盖最长的
+  month 窗口）。
+- 配额耗尽返回 **429 + `Retry-After`**（`rate_limit_exceeded`），消息里点名是哪个
+  模型用满了，客户端可等窗口重置后自动恢复；模型越权才是 403。
+  **admin 密钥永不受配额限制**（其 scope 上的配额也不强制），避免把管理员
+  锁在门外。
+- 窗口用量按整点小时分桶统计，实际释放比配置窗口最多晚 1 小时（配额宁可晚释放
+  也不超发）。
+- `PUT /api/keys/{key}` 提交 `models` 即同时提交它们的配额（配额就是 scope 的一部分，
+  不存在会与模型列表脱节的第二份预算）。显式 `0` = 解除该模型的限制。
+
+#### 配额拒绝 vs 容量拒绝：两种「拒绝」含义不同
+
+配额在**入口**检查，不占用任何上游槽位，因此是廉价拒绝；容量不足则要走
+`busyWait` 有界等待后才返回 503。实测（容量 8、0.6s/请求、100 并发）：
+
+| 情况 | 状态码 | 延迟 | 是否打到上游 |
+|---|---|---|---|
+| 配额用尽（`req_quota` 50 → 100 并发） | `429 rate_limit_exceeded` + `Retry-After` | **19–21ms** | 否 |
+| 容量打满（同一拓扑，未限配额） | `503 upstream_error` | ~2.6s（`busyWait`） | 超出部分否 |
+
+客户端因此可以区分「等窗口重置」与「等上游腾出容量」：前者只需等待，后者
+通常该降并发或换源。
+
+#### 配额的开销
+
+配额桶按 (密钥, 模型, 整点小时) 分桶保留 40 天，实测（AMD 7840HS）：
+
+- 窗口查询按窗口长度而非保留总量扫描：1h 窗口 49ns、24h 窗口 55ns、
+  30d 窗口 3.9μs（此前全扫保留总量，960 桶时 5.9μs）。
+- 记录一条请求：283ns、3 分配（与引入配额前相同，分配来自 ring buffer）。
+- 内存：生产形态（7 密钥 × 8 模型 × 2 源 × 满 40 天 retention）约 **3.7MB**。
+  按源 pin 的 `source::model` 桶**惰性创建**——只有当某条配额真的 pin 了
+  某个源时才维护，否则每条记录多写一份桶，在 20 密钥 × 8 模型 × 3 源下会
+  多占 18MB。
+  代价：配置 pinned 配额之前的历史用量无法事后按源拆分，所以 **pinned 配额的
+  首个窗口可能少算**。
+
+```bash
+# 配额耗尽时客户端看到（点名了具体模型）
+HTTP/1.1 429 Too Many Requests
+Retry-After: 2100
+{"error":{"type":"rate_limit_exceeded","message":"token quota exceeded for \"deepseek-v4-flash\" (5000000/5000000)"}}
+```
+

docs/deployment.md

@@ -0,0 +1,137 @@
+# 部署与迁移
+
+> 本文档面向运维：单机 systemd 部署、发行包安装、目录与权限、从 `llmsproxy`
+> 改名后的迁移与回滚。
+> 返回 [README](../README.md)。
+
+## 1. 单机部署（推荐）
+
+仓库根目录的 `deploy.sh` 负责：构建 → 预检配置 → 备份 → 原子替换二进制 →
+重启服务 → 后置校验。它**不**修改源码，也**不**假设全新机器。
+
+```bash
+sudo ./deploy.sh                      # 默认读取仓库内 config.yaml
+sudo ./deploy.sh --config other.yaml  # 投放并部署另一份配置
+```
+
+部署目标（改名后）：
+
+| 项 | 路径 |
+|---|---|
+| 二进制 | `/usr/local/bin/modelrouter` |
+| 配置 | `/etc/modelrouter/config.yaml` |
+| 适配器 | `/etc/modelrouter/adapters/` |
+| 插件（可选） | `/etc/modelrouter/plugins/`（需在配置里设 `plugin_dir`） |
+| 主密钥 | `/etc/modelrouter/master.key`（0600） |
+| systemd unit | `/etc/systemd/system/modelrouter.service` |
+
+`deploy.sh` 会在以下情况**中止**而不是"看起来成功"：
+
+- 部署后二进制起不来、或新配置 `-check` 不通过；
+- 凭据无法解封（`master.key` 丢失 / 与配置不匹配）；
+- 检测到改名前的 `/etc/llmsproxy`，但新配置尚未建立（否则会部署出一个
+  没有 key、没有源的空实例，所有请求都被拒）。
+
+## 2. 发行包安装
+
+三个产物由 `packaging/core-dist.sh` 生成到 `cmd/build/dist/`：
+
+```text
+modelrouter_<ver>_amd64.deb
+modelrouter-<ver>.x86_64.rpm
+modelrouter-<ver>-linux-amd64.tar.gz
+```
+
+deb / rpm 安装后：
+
+```bash
+sudo systemctl enable --now modelrouter
+sudo journalctl -u modelrouter -f
+```
+
+tar 包内只有：`modelrouter`、`config.example.yaml`、`modelrouter.service`、
+`migrate-to-modelrouter.sh`、`adapters/`。
+
+**发行包不含业务插件**（`channels` / `billing` / `quota` / `usage` 等）。
+它们保留在仓库中作为可选扩展，需要时手工放入 `plugin_dir`，网关不会自动安装、
+覆盖或删除。这是有意的：插件属于运维资产，生命周期不与内核绑定。
+
+首次启动会在 `/etc/modelrouter/config.yaml` 生成一份**随机 admin key** 的配置，
+key 打印在启动日志里。发行包不预置任何可用 key（历史上曾因此泄露过一个真实的
+`sk-gw-local-0001`）。
+
+## 3. 目录与权限
+
+- `/etc/modelrouter` 是唯一的读写根（systemd unit 用 `ProtectSystem=strict` +
+  `ReadWritePaths=/etc/modelrouter` 固化这一点）。
+- `master.key` 必须 0600。当前 unit 以 root 运行：降权到普通用户需要先把
+  `master.key` 交给该用户并统一目录属主，否则服务会"带着明文落盘继续跑"，
+  失败只留一行日志——这是独立变更，不要和加固混在一起。
+- `plugin_dir` 由运维维护；插件自带的状态文件（如 `.channels-store.json`）也在其中。
+
+## 4. 从 llmsproxy 迁移
+
+二进制、配置目录、服务名、环境变量已统一为 `modelrouter`：
+
+| 旧 | 新 |
+|---|---|
+| `/usr/local/bin/llmsproxy` | `/usr/local/bin/modelrouter` |
+| `/etc/llmsproxy/` | `/etc/modelrouter/` |
+| `llmsproxy.service` | `modelrouter.service` |
+| `LLMSPROXY_PLUGIN_DIR` | `MODELROUTER_PLUGIN_DIR` |
+| `LLMSPROXY_XDISPLAY_BASE` | `MODELROUTER_XDISPLAY_BASE` |
+| `LLMS_PROXY_MASTER_KEY` | `MODELROUTER_MASTER_KEY` |
+
+迁移脚本**默认 dry-run**，复制而非移动，旧目录永不删除：
+
+```bash
+sudo ./packaging/migrate-to-modelrouter.sh          # 预演（逐条打印将做什么）
+sudo ./packaging/migrate-to-modelrouter.sh --apply  # 执行
+```
+
+它做五件事：复制状态目录并改写配置里的旧绝对路径、复核 `master.key` 0600、
+安装新二进制（旧文件保留）、由旧 unit 生成新 unit（改写 ExecStart /
+WorkingDirectory / ReadWritePaths）、复制 `/usr/share` 资源；最后跑
+`modelrouter -check` 并 `systemctl daemon-reload`。
+
+确认新服务可用后，再手工清理旧布局（脚本不会删）：
+
+```bash
+sudo systemctl disable --now llmsproxy.service
+sudo rm -f /etc/systemd/system/llmsproxy.service
+sudo rm -rf /usr/local/bin/llmsproxy /etc/llmsproxy /usr/share/llmsproxy
+```
+
+回滚（不需要旧文件被删）：
+
+```bash
+sudo ./packaging/migrate-to-modelrouter.sh --rollback --apply
+```
+
+**读取侧兼容**：旧配置路径与旧环境变量仍被接受，所以"先换二进制、后改 unit"
+这种顺序不会立刻失效。但写入永远落在新路径，迁移完成后不要两台布局混用。
+
+## 5. 部署后验证清单
+
+```bash
+# 1) 进程与监听
+systemctl status modelrouter --no-pager
+ss -ltnp | grep modelrouter
+
+# 2) 鉴权生效（无 key 必须 401）
+curl -s -o /dev/null -w '%{http_code}\n' http://127.0.0.1:8080/v1/models
+
+# 3) 带 key 能列出模型
+curl -s -H "Authorization: Bearer $KEY" http://127.0.0.1:8080/v1/models | head -c 400
+
+# 4) 真实请求走通（证明 LuaJIT 适配器路径未被打断）
+curl -s -H "Authorization: Bearer $KEY" \
+  -d '{"model":"AUTO","messages":[{"role":"user","content":"hi"}]}' \
+  http://127.0.0.1:8080/v1/chat/completions | head -c 400
+
+# 5) 仓库文件与生产文件一致（部署脚本的收尾检查）
+sha256sum /usr/local/bin/modelrouter build/modelrouter
+```
+
+第 4 步失败但第 3 步成功，通常是适配器或上游凭据问题，不是部署问题：
+查看 `journalctl -u modelrouter -n 100`。

docs/development.md

@@ -0,0 +1,34 @@
+# 开发与目录结构
+
+> 本文档由 README 拆分而来：目录职责、开发协作、测试。
+> 返回 [README](../README.md)。
+
+## 目录
+
+```
+cmd/modelrouter            # 入口（纯 Go 服务端）
+cmd/gui                  # Electron 桌面 GUI（内嵌核心 + 托盘 + 打包）
+internal/config          # YAML 配置加载/校验
+internal/lua             # LuaJIT VM + worker 池 + AdapterCache + 内置适配器 (embed)
+internal/provider        # Provider(HTTP) + Registry(路由)
+internal/gateway         # OpenAI 兼容 HTTP 服务 + 鉴权 + SDK/流式 + WebUI
+internal/types           # 统一格式 & OpenAI wire 类型
+```
+
+## 开发协作
+
+- **Git 分支工作流**：主分支 / 特性分支 / 发布分支的规范见
+  **[docs/git-workflow.md](docs/git-workflow.md)**（[English](docs/git-workflow-en.md)）。
+  主线：`main` 永可部署 → 特性合入 `main` → 切 `release/vX.Y.Z` 打 tag 发布 →
+  hotfix 提交发布分支并 cherry-pick 回 `main`。
+- **Lua 适配器协议**：见 **[docs/lua-adapters.md](docs/lua-adapters.md)**
+  （[English](docs/lua-adapters-en.md)）。
+
+## 测试
+
+```bash
+go test -tags luajit ./...
+```
+
+覆盖：配置校验、适配器加载/变换、签名钩子、Gateway 鉴权、SDK 结算、SSE 流式、
+模型路由、多模态透传与 thinking 预算。

docs/features.md

@@ -0,0 +1,48 @@
+# 实现与完整特性
+
+> 本文档由 README 拆分而来：架构来源与完整特性清单（README 只保留摘要）。
+> 返回 [README](../README.md)。
+
+## 实现
+
+统一 OpenAI 兼容网关：把多个上游 LLM 源（DeepSeek、Qijiar、OpenAI、Anthropic、Gemini、
+Groq、Mistral、Ollama、KimiCode…）通过 **Lua 适配器** 做协议转换，对内网暴露
+一个标准 `OpenAI Chat Completions` 接口（`/v1/chat/completions` + `/v1/models`），
+支持单次结算、SSE 流式、**AUTO 模型路由**与**多模态**透传。
+
+从 [HomeAgent](https://gitcode.com/JianFeeeee/HomeAgent) 的多源 LLM 适配层
+（`internal/agent/api/provider.go` + `internal/lua/adapters/*`）抽离而来并独立演进。
+
+## 特性
+
+- **多源**：一个进程内配置任意多个上游源，按请求的 `model` 自动路由。
+- **AUTO 模式**：`default_model: AUTO` 时按优先级页保存的 AUTO 链档位逐档调度并发请求。
+- **统一输出**：所有源对外都是 OpenAI 格式（含 `reasoning_content`、`tool_calls`、`usage`）。
+- **生图**：`POST /v1/images/generations`，`kind: image` 的模型独立路由。
+- **多模态**：`content` 数组（`image_url` 等）在多源间无损透传，Anthropic/Gemini/Ollama 自动转换。
+- **LuaJIT VM**：基于 golua 绑定的 LuaJIT，每个适配器独立 VM + worker 池，安全并发。
+  池是 **弹性** 的：`Σ max_concurrent` 只是上限而非预分配，状态按需创建、空闲时自动回收，
+  因此空载网关几乎不持有 Lua 状态（启动 0 个）。
+  - 扩容步长由适配器最大并发决定：`clamp(ceil(max/8), 1, 8)`，且仅在**真实并发争用**
+    （所有现有状态均已占用）时批量预热，顺序流量始终只用 1 个状态。
+  - 缩容步长由当前连接数决定：`clamp(ceil(冗余/(1+占用)), 1, 冗余)`——无连接时一轮收到
+    常驻下限，忙时每轮只释放 1 个，保护热路径。WebUI「适配器」页展示实时池指标。
+- **thinking 预算**：请求里的推理开关/预算归一为中立的 `thinking:{enabled,budget_tokens}`，
+  适配器映射到各自协议（Anthropic `thinking.budget_tokens`、Gemini `thinkingConfig`、
+  DeepSeek `thinking.type="disabled"`）。
+- **Lua 适配协议**：每个源挂一个 `.lua` 适配器，完成 `transform_request` /
+  `transform_response` / `transform_stream_chunk` 双向转换，协议差异全在 Lua 层。
+- **签名 / 请求头钩子**：适配器可定义 `build_headers(meta)`，在 Go 发 HTTP 前
+  动态注入/签名请求头——用于云端 API 校验调用方 app（如 KimiCode 只放行特定
+  agent）。提供 `hmac_sha256_hex` / `sha256_hex` / `base64_encode` 等签名辅助。
+- **WebUI**：内置管理界面，可在线查看/新增/编辑上游源与模型、配置 AUTO 优先级链、
+  管理密钥模型范围，写入运行时文件持久化。
+- **密钥加密存储**：运行时文件中的上游 `api_key`、自定义请求头值、网关 key 均以
+  AES-256-GCM 加密落盘（`master.key` 独立 0600，或 `MODELROUTER_MASTER_KEY`）。
+- **实时源探测**：状态页探测各源可达性（`GET {base}/models`，失败回退最小请求），
+  不污染正常调度的退避状态，错误信息在 UI 可悬停查看。
+- **鉴权**：网关自身用 `gateway_keys` 校验客户端 Bearer key；与上游各自的 key 相互独立。
+- **流式**：SSE `chat.completion.chunk`，含角色首包与 `[DONE]` 收尾。
+- **全链路审计**：HTTP 访问日志、登录/配置变更审计、请求记录 CSV 导出、密钥用量统计 CSV 导出
+- **AES-256-GCM 加密存储**：运行时文件敏感字段加密落盘，master.key 0600 权限，支持环境变量注入主密钥
+

docs/gui.md

@@ -0,0 +1,86 @@
+# 桌面 GUI（Electron）
+
+> 本文档由 README 拆分而来：Headless 与 Desktop 两种产物、构建与运行。
+> 返回 [README](../README.md)。
+
+## 桌面 GUI（Electron）
+
+可选独立产物，面向内网桌面端用户：**内嵌 ModelRouter 核心**（桌面壳形态），
+一键启停、系统托盘常驻、开机自启、静默启动，内容区一比一内嵌完整 WebUI
+（状态/对话/密钥/优先级/源/适配器 六个页面，免登录）。服务器用户继续使用纯 Go 二进制。
+
+### 两种安装包：Headless 与 Desktop
+
+发布到 Release 的安装包分为两类，面向不同用户：
+
+- **Headless（服务器版）**：纯核心二进制（无 Electron），配置驱动，适合服务器、
+  容器、systemd 等无人值守场景。发布包含 `deb` + `rpm`（含 systemd unit、内存调优
+  与示例配置）与 `tar.gz`（二进制 + `config.example.yaml` + `adapters/` + README），
+  Windows 为 `.zip`。
+- **Desktop（桌面版）**：Electron GUI 安装包，内嵌核心，适合个人桌面日常使用。
+
+### 选择 GUI 还是纯后端
+
+| 场景 | 推荐 | 理由 |
+| ------ | ------ | ------ |
+| 服务器 / 内网网关 / 无人值守常驻 | **Headless**（单二进制） | 极轻量（二进制 ~8MB；RSS 随源数量，单源 ~10MB、16 源 ~32–35MB，见[内存占用](#内存占用实测与调优)），零依赖单进程，直接跑在 systemd/任意容器里，远程管理 |
+| 个人桌面日常使用 / 多设备内网共享 | **Desktop**（GUI） | 免登录内嵌 WebUI、系统托盘一键启停、开机自启、静默后台，适合不懂命令行的使用者 |
+| Windows 桌面 | **Desktop** | 纯后端在 Windows 上需自行注册服务，GUI 提供原生托盘/自启体验 |
+| CI 一键出三平台安装包 | **Desktop 打包脚本** | `make gui-*` 系列出 deb/AppImage/NSIS，可进发行流水线 |
+
+两者完全同源：GUI 内嵌的就是纯后端的同一份 `modelrouter` 核心（LuaJIT 版），
+配置/适配器格式完全一致，可随时互换。
+
+### 构建与运行
+
+```bash
+cd cmd/gui
+npm install
+make gui          # 或 npm run dev —— 开发运行（需先 make build 生成 bin/modelrouter）
+make gui-deb      # 本机构建 deb（分享给其它 Linux 用户）
+make gui-dist     # 构建 deb + rpm + AppImage（linux，rpm 需系统 rpmbuild）
+make core-dist    # 构建核心 deb + rpm + tar.gz（服务器版，需 nfpm）
+make gui-win      # 构建 Windows nsis 安装包（本机 mingw + wine）
+make gui-win-docker  # 构建 Windows nsis 安装包（全 docker 自足，含 wine）
+make gui-dist-dir # 仅解包产物（调试用，不产安装包）
+```
+
+产物在 `cmd/build/gui-dist/`：`ModelRouter-<ver>.AppImage`、
+`modelrouter-gui_<ver>_amd64.deb`、`ModelRouter Setup <ver>.exe`（win）；
+核心包在 `cmd/build/dist/`：`modelrouter_<ver>_amd64.deb`、`modelrouter-<ver>.x86_64.rpm`。
+rpm 需系统 `rpmbuild`。每次打包后都会跑 `packaging/verify-dist.sh` 产物体检
+（大小下限门禁——1.3.0 曾因宿主 wine 损坏产出 264KB 的残缺 exe 却没有被拦住）。
+发布时会把产物重命名为 `ModelRouter-Desktop-*`（GUI）与 `ModelRouter-Headless-*`（纯后端）
+两类上传到 GitCode Release。
+
+#### Windows 打包（docker 固化方案，推荐）
+
+`make gui-win-docker` 一步完成：交叉编译核心 + NSIS 打包**全在 docker 内**，宿主机
+**无需 mingw、无需 wine、无需 node**。镜像内置 wine（含 i386 运行时），修复了
+此前宿主 wine 损坏（缺 syswow64 → `c0000135`）导致产出 264KB 残缺 exe 的问题。
+
+- 构建镜像：`cmd/gui/docker/win-builder/Dockerfile`（`golang:1.25` + mingw-w64 +
+  node + wine32/wine64 + 预编译 LuaJIT windows 版 `lua51.dll` 与 import lib）。
+  LuaJIT 源码取自 gitcode 镜像（国内可达），github 兜底；Go 依赖走 `https://goproxy.cn`。
+- 一键发布：`cmd/gui/scripts/dist-win-docker.sh` —— 核心编译 + electron-builder
+  NSIS 全部在容器内执行（`USE_SYSTEM_WINE=true` 走镜像内的 wine）；electron-builder
+  的 ~400MB 工具链缓存挂到 docker volume，二次构建增量复用。
+- 首次运行自动构建 `modelrouter/win-builder` 镜像（约 5–10 分钟）；已存在则直接复用。
+- 打包完成后自动跑 `packaging/verify-dist.sh` 门禁：exe 低于 5MB / deb·rpm·7z 低于
+  10MB 即失败——残缺安装包不可能再溜进 Release。
+
+```bash
+# 全 Windows 发布（核心重编 + NSIS）：
+make gui-win-docker
+```
+
+### 特性
+
+- **内嵌核心**：自动拉起捆绑的 `modelrouter`（LuaJIT 版），配置/密钥/adapter 存于
+  用户数据目录 `<userData>/profile/`，首次运行自动生成随机 admin key 写入 `keys`
+  （非 seed）并自注入，WebUI 免登录、不会提示"更换初始密钥"。端口可在设置中更改（默认 8787）。
+- **系统托盘**：内核状态点、开机自启/静默启动开关、启停/重启内核、退出。
+- **静默启动**：`--silent` 参数或设置项，启动仅驻托盘不弹窗；首次运行始终显示窗口。
+- **开机自启**：Windows/macOS 用 `setLoginItemSettings`；Linux 写
+  `~/.config/autostart/modelrouter-gui.desktop`，静默开启时自动附加 `--silent`。
+

docs/performance.md

@@ -0,0 +1,60 @@
+# 性能与内存
+
+> 本文档由 README 拆分而来：实测内存占用、调优手段、thinking 预算。
+> 返回 [README](../README.md)。
+
+## 内存占用（实测与调优）
+
+内存不是一个定数：它随**配置的源数量**增长（每个源一个 `http.Transport` 与连接池、
+一组健康状态），与运行时长无关。本机实测（Linux x86_64，12 核）：
+
+| 部署形态 | 启动 RSS | 稳态 RSS |
+|---|---|---|
+| 1 源 / 1 适配器（最小配置） | ~4 MB | ~10 MB |
+| 1 源 + 29 MB 历史审计日志 | ~19 MB | ~20 MB |
+| **16 源 / 13 适配器 / 59 模型（本机生产）** | ~28 MB | **~32–35 MB** |
+
+> 历史参考：本项优化前同一生产配置为 **~105 MB**。降幅来自三处：审计日志的**原始记录**
+> 不再常驻内存（约 25 MB；统计聚合仍扫全量，但扫完即释放）、Lua 状态池不再单调增长、
+> 以及下面两个运行时开关。
+
+内存构成（生产实例分段测量，`/proc/<pid>/smaps`）：
+
+| 区域 | RSS | 说明 |
+|---|---|---|
+| Go 堆 | ~14 MB | provider/registry/scheduler 结构 + 连接池缓冲 |
+| 其他匹名（线程栈 / LuaJIT chunk / runtime） | ~12 MB | 与线程数、已加载适配器数相关 |
+| 二进制 text+rodata | ~8 MB | 映射的可执行文件页（只读、可被内核回收） |
+| Go runtime 预留 | ~4 MB | VSZ 上看到的 2 GB+ 是地址空间预留，不占物理内存 |
+| 共享库 | ~3 MB | libc / libluajit / libm |
+
+**两个推荐的部署开关**（只适用于环境变量，无需改代码）：
+
+```ini
+# /etc/systemd/system/modelrouter.service
+Environment=GOGC=50
+Environment=MALLOC_ARENA_MAX=2
+```
+
+- `MALLOC_ARENA_MAX=2`：LuaJIT 的分配走 cgo → glibc malloc，glibc 默认允许 `8×nproc`
+  个 per-thread arena，每个碰到 malloc 的 OS 线程会占用一个（各约 1 MB，且**不归还给系统**）。
+  实测从 8–15 个 arena（约 7–12 MB）降到 **0**。
+- `GOGC=50`：把 Go 堆增长目标减半。**单独使用无效**（省下的堆会立即被更多 glibc
+  arena 吃掉，实测 20.3 → 21.5 MB 反而变大），必须与 `MALLOC_ARENA_MAX` 配合，
+  两者同时开启才降 **~19%**。网关是 I/O 密集型（本机 9 小时仅消耗 1min10s CPU），
+  多出的 GC 周期与它的空闲 CPU 相比可忽略。
+
+> 桌面版（Electron）已在 `cmd/gui/main.js` 里默认为内嵌核心注入这两个开关；手动 export 同名
+> 环境变量可覆盖。如果你用自己的 systemd unit / 容器，建议照上面加上。
+
+可选：`GOMEMLIMIT=48MiB` 作软上限，实测再省 ~1 MB；代价是逐近上限时 GC 转激进，
+温和场景不必开。
+
+## thinking 预算
+
+请求体带 `"thinking": {"enabled": false}` 时关闭推理；`"thinking": {"enabled": true,
+"budget_tokens": 4096}` 指定预算。内核只定义这个**中立**语义字段，映射到具体协议由
+适配器负责（Anthropic `thinking.budget_tokens`、Gemini `thinkingConfig.thinkingBudget`、
+DeepSeek `thinking.type = "disabled"`）。内核不再向适配器泄漏 `disable_thinking` /
+`extra_body` 这类私有字段。
+

docs/plugins.md

@@ -15,6 +15,37 @@ ModelRouter 的插件是**单个 `.lua` 文件**，放在 `config.yaml` 的 `plu
 
 ---
 
+## 0. 插件 SDK：唯一的公开接口
+
+插件与网关之间**只有一个**公开接口表 `modelrouter`。它不是 Go 包的导出符号，
+而是内核在启动插件 VM 时注入的 Lua 表；插件通过它调用内核能力，永远不需要
+（也不应该）依赖任何 `__llmsproxy_*` 全局、Go 结构体或内部函数名。
+
+```lua
+modelrouter.sdk_version   -- 字符串，契约版本；SDK 形状变化才递增
+modelrouter.http(spec)    -- 受出站白名单约束的 HTTP 请求
+modelrouter.secrets       -- 本插件专用的凭据保险库（set/get/delete/keys）
+modelrouter.kernel        -- 内核动作面（actions/call，受运维授权约束）
+modelrouter.adapters      -- 安装/更新/删除本插件自己的适配器
+modelrouter.plugin_dir    -- 本插件所在目录（内核给出的 ground truth）
+```
+
+规则：
+
+- **只依赖上表**。内核重命名内部全局、改实现或加守卫时，插件不受影响；
+  这正是把六个散落的全局收进一张表的原因。
+- **能力可能缺失**。未授予的能力（例如出站 HTTP 全禁）在表里是 `nil` 而不是
+  "存在但一调用就报错"，插件应先判断再使用。
+- **`sdk_version` 用来做兼容分支**。增量字段不递增版本；字段删除或调用约定
+  变化才递增。
+- **历史别名**：`llmsproxy`（旧产品名）与 `__llmsproxy_{http,secrets,kernel,
+  adapters,plugin_dir}` 仍然指向同一批对象，只为已安装的插件继续可用而保留，
+  已废弃；新插件使用 `modelrouter.*`。
+- 环境变量：插件目录优先读 `MODELROUTER_PLUGIN_DIR`，回退旧名
+  `MODELROUTER_PLUGIN_DIR`。
+
+---
+
 ## 1. 快速上手
 
 一个最小的插件：
@@ -65,16 +96,14 @@ return plugin   -- 必须返回一个 table
 core.New
   └─ lua.NewVM(adapter_dir).Start()        适配器状态
   └─ lua.NewPlugins(vm, plugin_dir)
-       ├─ SeedBundled()                    仅当目录不存在时写入内置插件（目前是 billing）
-       └─ LoadDir()                        按文件名字典序逐个加载
+`plugin_dir` 中的 Lua 文件由网关逐个加载；安装包不携带业务插件，也不会自动创建或写入插件文件。首次启用时，先把需要的 `.lua` 插件放入该目录，再启动网关。
 ```
 
 **加载失败不影响网关启动。** 一个语法错误的插件会被记录在 `GET /api/plugins`
 的 `error` 字段里，永远不会被调用。这与适配器一致，但理由更强：插件是可选的
 第三方扩展，因为一个 `.lua` 打错字就让网关起不来是错误的取舍。
 
-**目录一旦存在就是权威的。** 与适配器同规则：首启会 seed 内置插件，之后目录里
-的文件说了算，删除或编辑内置插件都是真实生效的操作。
+插件目录由运维独立管理。网关不会 seed、覆盖或删除其中的插件；目录缺失时只会跳过插件加载，不影响网关启动。
 
 ### 2.1 热更新
 
@@ -294,7 +323,7 @@ PUT    /api/plugins/{name}/state     替换状态（admin）
     "loaded": true
   }],
   "hook_errors": { "request_end": { "count": 3, "last_error": "billing: ..." } },
-  "plugin_dir": "/etc/llmsproxy/plugins"
+  "plugin_dir": "/etc/modelrouter/plugins"
 }
 ```
 
@@ -428,7 +457,7 @@ Cookie 鉴权都自然生效，无需暴露额外端口。
 -- 每小时的定时任务：轮换 token 并重写适配器
 function plugin.refresh_token()
   local tok = fetch_new_token()          -- 插件自己的取 token 逻辑
-  local name, err = __llmsproxy_adapters.install(
+  local name, err = modelrouter.adapters.install(
     "myauth",
     'return { name = "myauth", endpoint = "/chat/completions",' ..
     ' headers = { ["Authorization"] = "Bearer ' .. tok .. '" } }'
@@ -439,9 +468,9 @@ end
 
 | 调用 | 参数 | 返回 |
 |---|---|---|
-| `__llmsproxy_adapters.install(name, source)` | 适配器名 + Lua 源码 | `(生效名, nil)` 或 `(nil, err)` |
-| `__llmsproxy_adapters.update(name, source)` | 同上（只限自己装的） | `(true, nil)` 或 `(nil, err)` |
-| `__llmsproxy_adapters.remove(name)` | 适配器名（只限自己装的） | `(true, nil)` 或 `(nil, err)` |
+| `modelrouter.adapters.install(name, source)` | 适配器名 + Lua 源码 | `(生效名, nil)` 或 `(nil, err)` |
+| `modelrouter.adapters.update(name, source)` | 同上（只限自己装的） | `(true, nil)` 或 `(nil, err)` |
+| `modelrouter.adapters.remove(name)` | 适配器名（只限自己装的） | `(true, nil)` 或 `(nil, err)` |
 
 出错时不抛异常，而是返回 `(nil, err)`——与钩子/HTTP 处理器同一套约定，
 一个手滑的插件错误不会带走网关。
@@ -494,24 +523,66 @@ end
 
 ### 6.4 命名
 
-- 全局 `__llmsproxy_plugin` 存放插件返回的 table，**不要占用**
+- 全局 `__llmsproxy_plugin` 是内核存放插件返回 table 的**内部位置**，不属于 SDK，**不要占用**
 - 插件间状态完全隔离，一个插件改不了另一个的全局
 
 ---
 
-## 7. 内置计费插件
+## 7. Billing 插件示例
 
-`plugins/billing.lua`，默认随首启 seed 进来，开箱可用。
+`internal/lua/plugins/billing.lua` 是仓库中的可选插件示例，不随核心安装包分发，也不会在启动时自动安装。需要使用计费功能时，应将该脚本单独复制到配置的 `plugin_dir` 后再重启网关。
 
-### 7.1 计费维度
+### 7.0 分层：billing 完全在插件里，内核没有任何 billing 代码
 
-| 维度 | 用途 | 优先级 |
+`billing` 是**插件自己的事**：规则文件、编译、计价、账目、规则编辑页都由
+`plugins/billing.lua` 负责。内核里没有 billing 的配置结构、端点、编译或迁移逻辑
+—— 它只提供所有插件通用的能力（加载 `.lua`、插件状态持久化、插件目录、
+`plugin.http` 路由、插件可用的 `io`/`os`）。
+
+| 项 | 位置 |
+|---|---|
+| 规则文件 | `<plugin_dir>/billing.json`（插件自己读写；内核不知道这个文件） |
+| 计费账目 | 插件状态文件 `<plugin_dir>/.billing.state.json`（内核的通用状态持久化） |
+| 规则编辑 API | `/api/plugins/http/billing/rules`（GET/PUT/POST/DELETE，插件路由） |
+| 编译/计价 | 插件内 `compileRules()` / `priceFor()` / `costFor()` |
+
+规则以**源名**（`source`）为键，而不是 URL：URL→源名的映射属于内核配置，插件
+不需要为了它和内核对接；编辑页在浏览器里，直接取 `/api/sources` 给运营选源名。
+`url: "*"` 只表示"全局模型价目表"（模型维度本来就不分源）。
+
+> 历史：早期 `billing:` 写在 `config.yaml`、编译在 Go 的 `internal/billing` 里。
+> GitCode issue #1 的根因正是"Go 与 Lua 各写一半覆盖语义"，组合起来把模型价
+> 静默清零。现在内核侧 billing 代码已全部删除，规则文件也由插件自持。
+
+### 7.1 计费维度### 7.1 计费维度
+
+| 维度 | 用途 | token 价覆盖顺序 |
 |---|---|---|
-| `sources` | 源的固定价（如按请求计费的转发服务） | 中 |
-| `models` | 模型的 per-token 价 | 高 |
+| `default` | 兜底价（通常全 0） | 最低 |
+| `models` | 模型的 per-token 价 | 中 |
+| `sources` | 源级价：固定价，或覆盖该源全部模型的 token 价 | 高 |
 | `keys` | 单个网关 key 的覆盖价 | **最高** |
 
-**token 价**优先级：`keys` > `models` > `default`。
+**token 价**优先级：`keys` > `sources` > `models` > `default`。
+（实现顺序就是"后写覆盖前写"：`default` → `models` → `sources` → `keys`；
+所以**源级 token 价会覆盖模型价**。这是有意的：运营要能给某个源整体改价。）
+
+⚠️ 正因为 `sources` 会覆盖 `models`：**`mode: token` 的规则不产生任何源条目**。
+规则未列出的模型会进入未定价统计（可见），而不是静默计 0 —— 这是 issue #1 / PR #2
+的收口语义。只有 `free` / `subscription`（"边际成本为零"）会写有意为之的
+`{prompt:0, completion:0}` 零价条目。
+
+规则写法：
+
+```json
+{"active":"current","profiles":[{"id":"current","default":"unpriced","rules":[
+  {"source":"localzen","mode":"free"},
+  {"source":"codearts","mode":"token",
+   "models":{"glm-5.3":{"prompt":"0.15","completion":"0.60"}}},
+  {"url":"*","mode":"token","models":{"some-model":{"prompt":"1.0","completion":"2.0"}}}
+]}]}
+```
+
 **`per_request` 固定价是叠加的**（不覆盖），所以一个生图模型可以既算 token 又收固定费。
 **`cache_discount`** 见 §7.7。
 
@@ -599,6 +670,17 @@ curl -H "Authorization: Bearer $ADMIN_KEY" \
 
 仪表盘上有 "Unpriced" 卡片。**这个数应该是 0**；不是 0 就去补价目。
 
+**profile 级兜底**：`profiles[].default` 可设为 `free` 或 `unpriced`（默认）。
+它是插件运行时的策略（编译期写入 `prices.default_mode`）：
+
+| `default` | 没有单独规则的源 | `unpriced_reqs` |
+|---|---|---|
+| `unpriced`（省略时） | 零价，计入未定价债务 | +1 |
+| `free` | 零价，视为**刻意免费** | 不增加 |
+
+显式 `mode: unpriced` 的源即使在 `default: free` 下也仍然计入（"只统计不收费"
+的声明不能被兜底策略吞掉）。
+
 注意"未定价"不等于"免费"：这些请求的 `requests` / token 数**照常计入**
 `total` 与各维度，只有金额是 0。
 
@@ -651,7 +733,7 @@ cached = cache_hit_tokens                   → 全价 × cache_discount
 插件的 `log(level, msg)` 输出到网关的日志：
 
 ```
-journalctl -u llmsproxy -f | grep -i plugin
+journalctl -u modelrouter -f | grep -i plugin
 ```
 
 ---

docs/routing.md

@@ -0,0 +1,111 @@
+# 路由与调度
+
+> 本文档由 README 拆分而来：模型路由、AUTO 链、冷却与半冷却、源与适配器、源模板。
+> 返回 [README](../README.md)。
+
+## 模型路由
+
+`/v1/chat/completions` 的 `model` 解析顺序：
+
+1. `source/model` 或 `source:model` 前缀 → 指定源；
+2. 精确匹配某个源的 `model`；
+3. 配置 `default_model: AUTO` 时 → 走优先级页保存的 AUTO 链（见下）；
+4. 否则回落（返回错误）。
+
+任何 OpenAI 客户端，只要 `model` 设为某个源的 `name/任意名`，即可锁定走该源；
+设为 `AUTO`（或网关配了 `default_model: AUTO`）即自动按优先级链选源。
+
+## AUTO 链（优先级页）
+
+AUTO 调度**只**由优先级页保存的规则决定，源配置里的 `models[].priority` 数字不再
+参与调度、也不再显示。**聊天与生图是两套独立的链**，在优先级页面通过「聊天 / 生图」
+按钮切换编辑：
+
+- **聊天链**（`runtime_file` 的 `auto` 字段）：服务 `POST /v1/chat/completions`
+  的 `model: AUTO` 请求；只接受 `kind: chat` 模型，配置在链里的生图槽位会被忽略。
+- **生图链**（`runtime_file` 的 `auto_image` 字段）：服务 `POST /v1/images/generations`
+  的 `model: AUTO` 请求；只接受 `kind: image` 模型。未配置生图链时，AUTO 生图回落
+  到「所有含生图模型的源、按注册顺序」的旧行为，兼容旧配置。
+
+每条规则 = 一个「槽位」：`{ model, source, tier, token_quota, period, hours }`。
+
+- `tier` 表示优先级档位：同一档的模型并排、共享该优先级；档位从上到下递减。
+- 同一模型可配置多个槽位（如 A 源低配 → B 源低配 → A 源高配 → B 源高配），按档位顺延。
+- `token_quota` > 0 时该槽位在重置周期内用满即顺延到下一槽位；`period` 支持
+  `hour` / `week` / `month` / `nhour`（配合 `hours`），空 = 不限。
+- 同 tier 内按「偏好分」排序，失败槽位冷却后自动跳过；冷却 / 配额耗尽 / hard
+  错误会顺延到下一 tier，全部失败时返回 503 并附上每档失败摘要。
+
+## 冷却与半冷却探测（自愈调度）
+
+槽位失败后进入指数退避冷却（5s → 10s → 20s …，上限 **5 分钟**）。冷却不是全时段
+硬闸：
+
+- **前半段完全静默**：刚失败的槽位不接任何流量，避免对故障上游持续施压。
+- **后半段放行 1 个探测**：窗口过半后，同一 `(源, 模型)` 最多允许 **一个** 请求作为
+  探测通过。探测槽数 = `clamp(max_concurrent / 10, 1, 2)`——`max_concurrent: 10`
+  的源恰好放行 1 个探测请求。
+- **探测排在最后**：同 tier 内探测候选永远排在健康候选之后，只有没有正常槽可用时
+  才承接真实流量，因此探测不会抢走可用容量。
+- **探测即真实请求**：成功则立即清零失败计数与冷却，槽位当场回到正常轮询——
+  上游恢复/额度恢复后不必等满整个冷却窗口。失败则重开一个新窗口，下次探测顺延到
+  新窗口的中点，不会连续重试。
+
+分类冷却策略：
+
+| 情况 | 冷却 | 说明 |
+|---|---|---|
+| 传输错误 / 5xx | 5s→10s→20s…，上限 5min | 指数退避 |
+| 401 / 403 凭据错误 | 10min | 换 key 后由下一次探测自动接回 |
+| 429 限流 | 30s 固定 | 「太快」不等于「坏了」，不进指数阶梯 |
+| 配额耗尽（`insufficient_quota` / `余额不足` 等） | 对齐配额窗口，上限 30min | **不进指数阶梯、不涨失败计数**，额度恢复即可用 |
+
+WebUI「优先级」页的槽位健康标签会区分 `冷却` / `待探测` / `探测中`，悬停可看窗口
+起点、探测放行时刻与完全恢复时刻。
+
+## 源（Source）与适配器（Adapter）的关系
+
+- **源** 是到某个上游的连接描述：`name`、`base_url`、`api_key`、模型列表与优先级。
+- **适配器** 是协议转换逻辑（Lua 脚本）：把统一 OpenAI 格式请求转成上游原生格式，
+  再把上游响应/流式分块转回统一格式。
+- 一个适配器可被多个源复用（如 `openai.lua` 同时服务多个 OpenAI 兼容站点）；
+  同一个源也可以切换不同适配器（改 `adapter` 字段即可）。
+- 源决定“连谁、暴露哪些模型”，适配器决定“怎么对话”——二者在 `sources[]` 条目中
+  通过 `adapter` 字段关联。
+
+加载流程（`internal/core` 的 `Core` 负责装配）：
+
+1. 启动时 `lua.NewVM(adapter_dir)` 加载全部适配器：内置适配器（编译期 embed）+
+   `adapter_dir` 下的同名覆盖文件；内置文件写在代码内，覆盖文件要求更高优先级。
+2. `config.Load` 读取 `config.yaml`，`config.NewStore(runtime_file)` 读 WebUI 改动的
+   运行时源，二者按名称合并成完整源列表。
+3. `rebuildRegistry` 为每个源创建 `provider.Provider`（持有目标适配器），并按源的
+   并发上限配置适配器 worker 池**上限**（不预分配，见下）。
+4. 请求进来时 `Registry` 按 `model` 路由到 Provider，Provider 调适配器
+   `transform_request` → HTTP 发送 → `transform_response` / `transform_stream_chunk`。
+
+WebUI 上的"新增/编辑源"、"上传 Lua 适配器"、"改 AUTO 优先级链"与"密钥模型范围"
+都即时生效（写入运行时文件或 `adapter_dir` 后重新装配，无需重启）；直接编辑
+`config.yaml` / `adapter_dir` 下的文件则需要重启进程才会重新加载。
+
+在 WebUI 删除源/适配器时：运行时（WebUI 创建）的源与 `.lua` 文件会真正移除；
+`config.yaml` 中定义的基源无法改写配置文件，采用删除标记隐藏（重启后仍隐藏），
+在 UI 里重新添加同名源即可恢复。
+
+## 源模板（多 key 负载均衡）
+
+多个上游 key 共享同一套配置（URL / 适配器 / 模型列表 / 并发 / rpm）时，逐一复制
+整段 source 太冗余。**源模板**保存除 `name` 和 `api_key` 以外的全部源字段，
+从模板一键创建多个仅 key 不同的源，天然支持多 key 负载均衡（每个展开源独立调度、
+独立冷却、独立健康状态）。
+
+- WebUI「源」页右上角「**模板管理**」按钮：列出全部模板，可编辑 / 删除 / 新建。
+- 添加源弹窗右上角两个按钮：
+  - **从模板创建** → 选模板 → 表单自动填充（name + key 仍需你填）；
+  - **存为模板** → 把当前表单非 key/name 字段存为新模板。
+- 模板持久化在 `runtime_file` 的 `source_templates` 字段，与运行时源一样热管理，
+  改完即时生效。模板本身不含密钥，可以安全共享 / 纳管。
+
+模板不会自己变成源——它只是「配方」。真正承担流量的是从模板创建出来的、带 key
+的具体源。
+

docs/webui.md

@@ -0,0 +1,23 @@
+# WebUI
+
+> 本文档由 README 拆分而来。
+> 返回 [README](../README.md)。
+
+## WebUI
+
+内置管理界面（`GET /`），登录后可在浏览器完成：
+
+- **状态页**：源在线状态（实时探测 + 悬停看错误）、模型/源/key 用量统计、请求记录，
+  支持按时间范围导出 CSV；点击模型可生成 pin 到该模型的连接配置。
+- **对话页**：流式/非流式调试。
+- **密钥页**：创建/编辑网关 key，为每个 key 配模型范围（模型 + 源 + token 配额 + 周期），
+  管理员管理全部 key，用户只看到自己的 key。每个模型砖块显示自己的配额徽标
+  （如 `1.0K 77×·1h`，未设配额显示 `∞`），点开可编辑该模型的 token 配额、
+  请求数配额与重置周期 —— 配额按模型独立生效，一个用满不影响同一 key 的其它模型。
+  「我的密钥」页对用户展示本 key 的模型范围与各自配额。
+- **优先级页**：拖拽积木配置 AUTO 链档位。
+- **源页**：在线增删改上游源（API key 等敏感字段加密落盘）。
+- **适配器页**：上传 / 删除 Lua 适配器脚本。
+
+改动写入 `runtime_file`（重启仍生效）。
+

e2e/e2e_test.go

@@ -1,4 +1,4 @@
-// Package e2e runs the real llmsproxy binary against mocked upstreams over
+// Package e2e runs the real modelrouter binary against mocked upstreams over
 // real HTTP: startup, auth, chat, streaming SSE, image generation, and
 // AUTO failover when a higher-priority source fails.
 package e2e
@@ -100,15 +100,15 @@ type gatewayUnderTest struct {
 func buildBinary(t *testing.T) string {
 	t.Helper()
 	dir := t.TempDir()
-	bin := filepath.Join(dir, "llmsproxy")
+	bin := filepath.Join(dir, "modelrouter")
 	if runtime.GOOS == "windows" {
 		bin += ".exe" // go build writes the exact name; Windows needs the suffix to exec
 	}
-	out, err := exec.Command("go", "build", "-tags", "luajit", "-o", bin, "llmsproxy/cmd/llmsproxy").CombinedOutput()
+	out, err := exec.Command("go", "build", "-tags", "luajit", "-o", bin, "modelrouter/cmd/modelrouter").CombinedOutput()
 	if err != nil && runtime.GOOS == "windows" {
 		// local Windows dev box may lack LuaJIT: fall back to the bundled
 		// Lua runtime (the e2e adapters used here are passthrough-only)
-		out, err = exec.Command("go", "build", "-o", bin, "llmsproxy/cmd/llmsproxy").CombinedOutput()
+		out, err = exec.Command("go", "build", "-o", bin, "modelrouter/cmd/modelrouter").CombinedOutput()
 	}
 	if err != nil {
 		t.Fatalf("build: %v\n%s", err, out)

go.mod

@@ -1,4 +1,4 @@
-module llmsproxy
+module modelrouter
 
 go 1.25.0
 

internal/billing/compile.go

@@ -1,323 +0,0 @@
-package billing
-
-import (
-	"fmt"
-	"strings"
-
-	"llmsproxy/internal/config"
-)
-
-// Compile turns one billing profile into the prices table the billing plugin
-// expects, resolving URL rules against the gateway's actual sources.
-//
-// WHY URL RULES NEED RESOLVING AT ALL: the operator declares pricing by URL
-// because that is what a provider's price list is keyed on, and because several
-// sources can point at the same URL. The plugin, however, looks up by
-// `payload.source` (a source NAME) and by model — it has no idea what URL a
-// request went to. So the URL match happens here, at compile time, where the
-// config's name->base_url mapping is known, and the result is expressed in the
-// dimensions the plugin already supports.
-//
-// Profiles are the "let the user choose" axis: the same URL can appear in
-// several profiles and switching recomputes this table, so a gateway can be
-// repriced without editing the plugin.
-func Compile(profile *config.BillingProfile, sources []config.Source) (map[string]interface{}, error) {
-	return CompileOpts(profile, sources, false)
-}
-
-// CompileOpts is Compile with an explicit policy for rules that match no
-// source.
-//
-// The default is STRICT and that is correct for startup: a profile with a
-// typo'd URL prices nothing, every request on it is recorded as unpriced, and
-// the bill silently comes out wrong. Failing the load is the right response.
-//
-// The rules editor needs the opposite. An operator editing a half-finished
-// profile — adding a rule before the source that will use it exists, or
-// renaming a URL while typing — must be able to save and then see the warning,
-// not be blocked by an error they can only resolve by guessing. Lenient mode
-// keeps the offending rule in the config and drops it from the compiled table,
-// so it is preserved as text and clearly reported, without pretending it
-// prices anything.
-func CompileOpts(profile *config.BillingProfile, sources []config.Source, lenient bool) (map[string]interface{}, error) {
-	if profile == nil {
-		return nil, fmt.Errorf("no billing profile")
-	}
-	prices := map[string]interface{}{
-		"currency": "USD",
-		"default":  map[string]interface{}{"prompt": 0.0, "completion": 0.0, "per_request": 0.0},
-		"sources":  map[string]interface{}{},
-		"models":   map[string]interface{}{},
-		"keys":     map[string]interface{}{},
-	}
-	if profile.Default != "" {
-		prices["default_mode"] = profile.Default
-	}
-
-	// Currency is a profile-level statement; a rule may override it.
-	for i := range profile.Rules {
-		if c := profile.Rules[i].Currency; c != "" {
-			prices["currency"] = c
-			break
-		}
-	}
-
-	matched := map[string]bool{}
-	for i := range profile.Rules {
-		rule := &profile.Rules[i]
-		targets := matchSources(rule.URL, sources)
-		if len(targets) == 0 && rule.URL != "*" {
-			// A rule for a URL no source uses is almost always a typo (or a
-			// source that was removed). Failing loudly beats a profile that
-			// silently prices nothing — the whole reason this is a config file
-			// instead of a hand-written JSON blob.
-			if lenient {
-				continue // kept in the config text, reported as a warning
-			}
-			return nil, fmt.Errorf("rule url %q matches no configured source base_url", rule.URL)
-		}
-		for _, srcName := range targets {
-			if matched[srcName] {
-				// First rule wins. Two rules matching one source is ambiguous,
-				// and silently letting the later one win makes the file's
-				// meaning depend on ordering the operator cannot see.
-				continue
-			}
-			matched[srcName] = true
-			if err := applyRule(prices, rule, srcName, sources); err != nil {
-				return nil, err
-			}
-		}
-	}
-
-	// Sources no rule matched fall through to the profile default, which the
-	// plugin applies via prices.default. Recording them explicitly means the
-	// UI can say "this source is unpriced" instead of leaving the operator to
-	// infer it from a zero.
-	var unmatched []string
-	for _, src := range sources {
-		if !matched[src.Name] {
-			unmatched = append(unmatched, src.Name)
-		}
-	}
-	if len(unmatched) > 0 {
-		prices["unmatched_sources"] = unmatched
-	}
-	return prices, nil
-}
-
-// matchSources returns the source names whose base_url matches pattern.
-// "*" matches every source (used as a catch-all default rule).
-func matchSources(pattern string, sources []config.Source) []string {
-	var out []string
-	if pattern == "*" {
-		for _, s := range sources {
-			out = append(out, s.Name)
-		}
-		return out
-	}
-	want := normalizeURL(pattern)
-	for _, s := range sources {
-		if normalizeURL(s.BaseURL) == want {
-			out = append(out, s.Name)
-		}
-	}
-	return out
-}
-
-// normalizeURL compares URLs the way an operator expects: trailing slashes and
-// case in the host are differences the provider's price list does not care
-// about, and requiring an exact byte match would make the config brittle.
-func normalizeURL(u string) string {
-	u = strings.TrimSpace(u)
-	u = strings.TrimRight(u, "/")
-	return strings.ToLower(u)
-}
-
-// applyRule writes one rule's pricing for one source into the prices table.
-func applyRule(prices map[string]interface{}, rule *config.BillingRule, srcName string, sources []config.Source) error {
-	switch rule.Mode {
-	case "free":
-		// Explicitly priced at zero. This is NOT the same as unpriced: a source
-		// the operator says is free must not appear in the unpriced warnings,
-		// or those warnings become noise and stop being read.
-		prices["sources"].(map[string]interface{})[srcName] = map[string]interface{}{
-			"prompt": 0.0, "completion": 0.0, "per_request": 0.0,
-		}
-		return nil
-
-	case "subscription":
-		// A fixed monthly commitment: the per-request MARGINAL cost is zero, and
-		// the flat fee is reported separately. Spreading a monthly fee across
-		// requests would invent a per-request number the provider never charges,
-		// and it would change every time traffic did.
-		prices["sources"].(map[string]interface{})[srcName] = map[string]interface{}{
-			"prompt": 0.0, "completion": 0.0, "per_request": 0.0,
-		}
-		fixed, _ := prices["subscriptions"].(map[string]interface{})
-		if fixed == nil {
-			fixed = map[string]interface{}{}
-			prices["subscriptions"] = fixed
-		}
-		cur := "USD"
-		if rule.Currency != "" {
-			cur = rule.Currency
-		}
-		fixed[srcName] = map[string]interface{}{
-			"monthly": rule.Subscription, "currency": cur,
-		}
-		return nil
-
-	case "unpriced":
-		// Deliberately left out of `sources` so the plugin's unpriced_models /
-		// unpriced_reqs counters catch it. That is the point: a subscription
-		// plan whose credits cannot be converted to tokens must be VISIBLE as
-		// unbilled, not quietly estimated.
-		return nil
-
-	case "token":
-		if len(rule.Models) == 0 {
-			return fmt.Errorf("rule for url %q: mode token requires at least one model", rule.URL)
-		}
-		models := prices["models"].(map[string]interface{})
-		for modelID, t := range rule.Models {
-			entry := map[string]interface{}{}
-			prompt, err := usdPerM(t.Prompt)
-			if err != nil {
-				return fmt.Errorf("model %q prompt: %w", modelID, err)
-			}
-			completion, err := usdPerM(t.Completion)
-			if err != nil {
-				return fmt.Errorf("model %q completion: %w", modelID, err)
-			}
-			entry["prompt"] = prompt
-			entry["completion"] = completion
-			if t.CacheDiscount != nil {
-				entry["cache_discount"] = *t.CacheDiscount
-			}
-			// reasoning is optional and stays ABSENT (not nil, not 0) when the
-			// rule does not set it, which is what makes the billing plugin skip
-			// the repricing branch entirely. Writing an explicit 0 here would
-			// mean "thinking is free" and silently under-bill every request
-			// from a provider that bills it like ordinary output.
-			if strings.TrimSpace(t.Reasoning) != "" {
-				rp, err := usdPerM(t.Reasoning)
-				if err != nil {
-					return fmt.Errorf("model %q reasoning: %w", modelID, err)
-				}
-				entry["reasoning_price"] = rp
-			}
-			if lc := t.LongContext; lc != nil {
-				if lc.Above <= 0 {
-					return fmt.Errorf("model %q long_context: above must be > 0", modelID)
-				}
-				above, err := longContextRate(lc, prompt)
-				if err != nil {
-					return fmt.Errorf("model %q long_context: %w", modelID, err)
-				}
-				entry["long_context"] = map[string]interface{}{
-					"above": lc.Above,
-					// Marginal, not blended. The plugin charges the tokens past the
-					// threshold at this rate and the rest at the base rate, so the
-					// threshold composes with a cache discount (which applies to the
-					// cached portion) instead of forcing the plugin to guess how
-					// the two interact.
-					"prompt": above,
-				}
-			}
-			if rule.Peak != nil {
-				entry["peak"] = compilePeak(rule.Peak)
-			}
-			models[modelID] = entry
-		}
-		// A specific-URL token rule needs a source entry so a model served
-		// from THIS url but absent from `rule.Models` is still recognised as
-		// belonging to a priced provider.
-		//
-		// A WILDCARD rule ("*") must NOT get one. "*" is a model catalogue — it
-		// says "these model ids have these prices, wherever they come from" —
-		// and marking every source priced at {0,0} would make priceFor() set
-		// priced=true for every model the catalogue does NOT list. Paid
-		// sources (commandcode, alittokenplan, …) would then record $0 instead
-		// of showing up in unpriced_reqs, which is precisely the silent
-		// zero-billing failure this whole plugin is built to make visible.
-		// The model lookup p.models[payload.model] already marks a listed
-		// model priced on its own, so the wildcard loses nothing.
-		if rule.URL != "*" {
-			if _, ok := prices["sources"].(map[string]interface{})[srcName]; !ok {
-				prices["sources"].(map[string]interface{})[srcName] = map[string]interface{}{
-					"prompt": 0.0, "completion": 0.0, "per_request": 0.0,
-				}
-			}
-		}
-		return nil
-	}
-	return fmt.Errorf("unknown mode %q", rule.Mode)
-}
-
-// usdPerM converts a USD-per-million string into the per-token rate the plugin
-// expects. Delegates to config.ParseFloatUSDPerM so the DSL validator and the
-// compiler agree byte-for-byte on what a valid price is — a value that passes
-// Validate() but fails here (or vice versa) would be the worst kind of drift.
-func usdPerM(s string) (float64, error) {
-	return config.ParseFloatUSDPerM(s)
-}
-
-// compilePeak renders the peak window in the shape the plugin reads:
-// { multiplier, windows = { { days = {...}, hours = { {lo,hi}, ... } } } }.
-//
-// The plugin reads `days` / `hours` pairs. The first version of the deployed
-// price table used { start, end, weekdays } — a shape nothing reads — so peak
-// traffic was billed at off-peak rates with no error anywhere. Compiling from
-// typed config fields removes the chance of writing the wrong key names by hand.
-func compilePeak(p *config.BillingPeak) map[string]interface{} {
-	win := map[string]interface{}{}
-	if len(p.Weekdays) > 0 {
-		days := make([]interface{}, 0, len(p.Weekdays))
-		for _, d := range p.Weekdays {
-			days = append(days, d)
-		}
-		win["days"] = days
-	}
-	if len(p.Hours) > 0 {
-		hours := make([]interface{}, 0, len(p.Hours))
-		for _, h := range p.Hours {
-			hours = append(hours, []interface{}{h[0], h[1]})
-		}
-		win["hours"] = hours
-	}
-	out := map[string]interface{}{"multiplier": p.Multiplier}
-	if len(win) > 0 {
-		out["windows"] = []interface{}{win}
-	}
-	return out
-}
-
-// longContextRate resolves the marginal per-token rate for the tokens ABOVE a
-// threshold, given the model's base per-million prompt price.
-//
-// Two ways to express it are supported because providers publish both: an
-// absolute rate (OpenAI quotes \$3.00/M past 200k) and a multiple (DeepSeek
-// quotes exactly 2x past 128k). Neither may be missing: a step with no rate is a
-// silent no-op, and an operator who wrote one meant something.
-func longContextRate(lc *config.BillingLongContext, basePerToken float64) (float64, error) {
-	hasPrompt := strings.TrimSpace(lc.Prompt) != ""
-	hasMult := lc.Multiplier > 0
-	switch {
-	case hasPrompt && hasMult:
-		p, err := usdPerM(lc.Prompt)
-		if err != nil {
-			return 0, err
-		}
-		// Both given: the absolute rate wins, because it is the one the provider
-		// actually quotes. Silently multiplying instead would produce a number
-		// the operator never wrote.
-		return p, nil
-	case hasPrompt:
-		return usdPerM(lc.Prompt)
-	case hasMult:
-		return basePerToken * lc.Multiplier, nil
-	default:
-		return 0, fmt.Errorf("needs either prompt or multiplier")
-	}
-}

internal/billing/wildcard_test.go

@@ -1,214 +0,0 @@
-package billing
-
-import (
-	"testing"
-
-	"llmsproxy/internal/config"
-)
-
-// The wildcard token rule must not blanket-mark every source as priced.
-//
-// Why this is a hazard and not a style question: priceFor() sets priced=true as
-// soon as the request's SOURCE has an entry in the price table, and that flag
-// is what keeps a request OUT of unpriced_reqs. A "*" rule is a model
-// catalogue — it prices specific model ids wherever they come from — but
-// compiling it into a {0,0,0} entry on all 21 sources made every model the
-// catalogue does not list look "priced at zero". Paid providers (commandcode,
-// alittokenplan, …) then stopped appearing in the unpriced warnings, which is
-// exactly the silent zero-billing failure the plugin exists to surface.
-func TestWildcardTokenRuleDoesNotMarkEverySourcePriced(t *testing.T) {
-	sources := []config.Source{
-		{Name: "free1", BaseURL: "https://free1.example.com/v1"},
-		{Name: "paid1", BaseURL: "https://paid1.example.com/v1"},
-		{Name: "paid2", BaseURL: "https://paid2.example.com/v1"},
-	}
-	profile := &config.BillingProfile{
-		ID: "p",
-		Rules: []config.BillingRule{
-			{URL: "https://free1.example.com/v1", Mode: "free"},
-			{
-				URL: "*", Mode: "token",
-				Models: map[string]config.BillingToken{
-					"only-catalogued-model": {Prompt: "1.00", Completion: "2.00"},
-				},
-			},
-		},
-	}
-	prices, err := Compile(profile, sources)
-	if err != nil {
-		t.Fatalf("compile: %v", err)
-	}
-	srcs := prices["sources"].(map[string]interface{})
-
-	// The explicitly free source keeps its entry (free is a deliberate 0).
-	if _, ok := srcs["free1"]; !ok {
-		t.Error("the explicit free rule lost its source entry")
-	}
-	// Paid sources must NOT be marked, or their unlisted models bill $0.
-	for _, name := range []string{"paid1", "paid2"} {
-		if _, ok := srcs[name]; ok {
-			t.Errorf("wildcard token rule marked %q as priced at 0 — its unlisted "+
-				"models will bill $0 instead of appearing as unpriced", name)
-		}
-	}
-	// The catalogue itself must still be present, or the rule prices nothing.
-	models := prices["models"].(map[string]interface{})
-	if _, ok := models["only-catalogued-model"]; !ok {
-		t.Error("the wildcard rule did not publish its model price")
-	}
-}
-
-// The inverse guard: a token rule for a SPECIFIC url still marks that source,
-// because it genuinely declares pricing for the models served from it.
-func TestSpecificURLTokenRuleStillMarksItsSource(t *testing.T) {
-	sources := []config.Source{
-		{Name: "mine", BaseURL: "https://mine.example.com/v1"},
-		{Name: "other", BaseURL: "https://other.example.com/v1"},
-	}
-	profile := &config.BillingProfile{
-		ID: "p",
-		Rules: []config.BillingRule{{
-			URL: "https://mine.example.com/v1", Mode: "token",
-			Models: map[string]config.BillingToken{
-				"m": {Prompt: "1.00", Completion: "2.00"},
-			},
-		}},
-	}
-	prices, err := Compile(profile, sources)
-	if err != nil {
-		t.Fatalf("compile: %v", err)
-	}
-	srcs := prices["sources"].(map[string]interface{})
-	if _, ok := srcs["mine"]; !ok {
-		t.Error("a specific-url token rule must still mark its own source priced")
-	}
-	if _, ok := srcs["other"]; ok {
-		t.Error("an unrelated source was marked priced by a specific-url rule")
-	}
-}
-
-// TestReasoningPriceIsOmittedUnlessSet pins the absence convention for the
-// thinking-token rate.
-//
-// A rule that says nothing about reasoning must produce a price entry with NO
-// reasoning_price key, so the billing plugin skips the repricing branch and
-// bills thinking at the ordinary output rate — which is what every provider
-// actually charges. Writing an explicit 0 instead would read as "thinking is
-// free" and silently under-bill every request from that model.
-func TestReasoningPriceIsOmittedUnlessSet(t *testing.T) {
-	sources := []config.Source{{Name: "up", BaseURL: "https://up.example/v1"}}
-
-	// Case 1: no reasoning declared → the key must not appear.
-	p1 := &config.BillingProfile{ID: "p", Rules: []config.BillingRule{{
-		URL: "https://up.example/v1", Mode: "token",
-		Models: map[string]config.BillingToken{
-			"m": {Prompt: "1.00", Completion: "2.00"},
-		},
-	}}}
-	prices, err := Compile(p1, sources)
-	if err != nil {
-		t.Fatal(err)
-	}
-	entry := prices["models"].(map[string]interface{})["m"].(map[string]interface{})
-	if _, present := entry["reasoning_price"]; present {
-		t.Errorf("reasoning_price present (%v) without a rule; the plugin would treat 0 as free",
-			entry["reasoning_price"])
-	}
-
-	// Case 2: declared → present, and converted to per-token like every other rate.
-	p2 := &config.BillingProfile{ID: "p", Rules: []config.BillingRule{{
-		URL: "https://up.example/v1", Mode: "token",
-		Models: map[string]config.BillingToken{
-			"m": {Prompt: "1.00", Completion: "2.00", Reasoning: "6.00"},
-		},
-	}}}
-	prices2, err := Compile(p2, sources)
-	if err != nil {
-		t.Fatal(err)
-	}
-	e2 := prices2["models"].(map[string]interface{})["m"].(map[string]interface{})
-	rp, ok := e2["reasoning_price"].(float64)
-	if !ok {
-		t.Fatalf("reasoning_price missing or wrong type: %#v", e2["reasoning_price"])
-	}
-	if rp != 6e-6 {
-		t.Errorf("reasoning_price = %v, want 6e-6 (USD per TOKEN)", rp)
-	}
-	// A malformed rate must be rejected, not silently ignored: an operator who
-	// typed "6 USD" expects either that value or an error, never a rule that
-	// looks applied and bills nothing.
-	p3 := &config.BillingProfile{ID: "p", Rules: []config.BillingRule{{
-		URL: "https://up.example/v1", Mode: "token",
-		Models: map[string]config.BillingToken{
-			"m": {Prompt: "1.00", Completion: "2.00", Reasoning: "six dollars"},
-		},
-	}}}
-	if _, err := Compile(p3, sources); err == nil {
-		t.Error("a non-numeric reasoning rate must fail the compile")
-	}
-}
-
-// TestLongContextStepCompiles pins the long-context step's two accepted forms
-// and its rejection of a step that would silently do nothing.
-//
-// Providers publish both shapes — DeepSeek says "2x past 128k", OpenAI quotes
-// an absolute "$3.00/M past 200k" — so both must compile. Neither may be
-// missing: a step with no rate is a rule that LOOKS like it handles long
-// context and does not, which under-bills without any visible sign.
-func TestLongContextStepCompiles(t *testing.T) {
-	sources := []config.Source{{Name: "up", BaseURL: "https://up.example/v1"}}
-
-	compile := func(lc config.BillingLongContext) (map[string]interface{}, error) {
-		p := &config.BillingProfile{ID: "p", Rules: []config.BillingRule{{
-			URL: "https://up.example/v1", Mode: "token",
-			Models: map[string]config.BillingToken{
-				"m": {Prompt: "1.00", Completion: "2.00", LongContext: &lc},
-			},
-		}}}
-		prices, err := Compile(p, sources)
-		if err != nil {
-			return nil, err
-		}
-		e := prices["models"].(map[string]interface{})["m"].(map[string]interface{})
-		return e["long_context"].(map[string]interface{}), nil
-	}
-
-	// Absolute rate (OpenAI's published shape).
-	lc1, err := compile(config.BillingLongContext{Above: 200000, Prompt: "3.00"})
-	if err != nil {
-		t.Fatalf("absolute long_context rate rejected: %v", err)
-	}
-	if lc1["above"] != 200000 || lc1["prompt"] != 3e-6 {
-		t.Errorf("absolute step = %#v, want above=200000 prompt=3e-6", lc1)
-	}
-
-	// Multiplier (DeepSeek's shape): 2x of the 1.00/M base.
-	lc2, err := compile(config.BillingLongContext{Above: 128000, Multiplier: 2})
-	if err != nil {
-		t.Fatalf("multiplier long_context rejected: %v", err)
-	}
-	if lc2["prompt"] != 2e-6 {
-		t.Errorf("multiplier step = %#v, want prompt=2e-6", lc2)
-	}
-
-	// Both given: the absolute rate wins, because that is the number the
-	// provider quotes. Multiplying instead would produce a figure the operator
-	// never wrote.
-	lc3, err := compile(config.BillingLongContext{Above: 128000, Prompt: "5.00", Multiplier: 2})
-	if err != nil {
-		t.Fatal(err)
-	}
-	if lc3["prompt"] != 5e-6 {
-		t.Errorf("both given = %#v, want the quoted absolute rate 5e-6", lc3)
-	}
-
-	// Neither: a no-op step must fail, not compile into a rule that silently
-	// bills long context at the base rate.
-	if _, err := compile(config.BillingLongContext{Above: 128000}); err == nil {
-		t.Error("a step with neither prompt nor multiplier must be rejected")
-	}
-	// A zero threshold would make every request "long".
-	if _, err := compile(config.BillingLongContext{Above: 0, Multiplier: 2}); err == nil {
-		t.Error("above=0 must be rejected")
-	}
-}

internal/config/config.go

@@ -7,7 +7,6 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"strconv"
 	"strings"
 	"time"
 
@@ -30,15 +29,6 @@ type Config struct {
 	Auto          []ModelScope `yaml:"auto,omitempty"`       // AUTO 调度链规则（WebUI 优先级页编辑，chat）
 	AutoImage     []ModelScope `yaml:"auto_image,omitempty"` // AUTO 生图调度链规则（WebUI 优先级页·生图）
 	Keys          []GWKey      `yaml:"keys,omitempty"`       // 网关密钥（WebUI 密钥页管理）
-	// BillingDSL declares per-URL pricing profiles for the billing plugin.
-	//
-	// Declarative, because hand-writing the prices JSON was error-prone in ways
-	// that looked successful: the first PUT went out without the required
-	// `prices` wrapper and silently REPLACED the accumulated totals with the
-	// price table; the second used a peak-window shape the plugin does not read,
-	// so peak traffic was billed at off-peak rates — no error, the bill just
-	// quietly halved. A schema-checked config file catches both at load.
-	BillingDSL *BillingDSL `yaml:"billing,omitempty"`
 	// PluginVisibility overrides the audience a plugin declared for itself,
 	// keyed by plugin name: "admin" hides a plugin (and its state) from user
 	// keys, "all" widens one that shipped admin-only, "user,admin" names roles
@@ -95,7 +85,14 @@ type Source struct {
 	// ProxyURL routes this source's HTTP(S) traffic through an explicit proxy
 	// (e.g. http://127.0.0.1:7890). Empty = direct connection. Needed for
 	// upstreams behind geo/IP blocks that only respond through a proxy.
-	ProxyURL      string                 `yaml:"proxy_url" json:"proxy_url,omitempty"`
+	ProxyURL string `yaml:"proxy_url" json:"proxy_url,omitempty"`
+	// Retry is an OPTIONAL upstream retry policy. It exists for upstreams that
+	// answer a business-level rejection with a status code the kernel would
+	// otherwise surface to the client — e.g. a per-account concurrency cap that
+	// clears by itself after tens of seconds. The kernel stays vendor-agnostic:
+	// the matcher is pure DATA (status list + response-body substrings), so no
+	// provider name ever appears in core.
+	Retry         *RetryPolicy           `yaml:"retry" json:"retry,omitempty"`
 	Meta          map[string]interface{} `yaml:"meta" json:"meta,omitempty"`
 	Temperature   float64                `yaml:"temperature" json:"temperature,omitempty"`
 	MaxTokens     int                    `yaml:"max_tokens" json:"max_tokens,omitempty"`
@@ -105,9 +102,66 @@ type Source struct {
 	RPM           int                    `yaml:"rpm" json:"rpm,omitempty"` // optional requests-per-minute cap (0 = unlimited)
 }
 
+// RetryPolicy describes when a failed upstream attempt may be retried BEFORE
+// anything is returned to the client. All conditions are data:
+//
+//	OnStatus       - retry only when the HTTP status is one of these
+//	OnBodyContains - ... AND the response body contains one of these substrings
+//	                 (empty = status alone decides)
+//	Delay          - wait between attempts ("10s" etc.); parsed on demand
+//	MaxAttempts    - TOTAL attempts including the first (1 or 0 = no retry)
+//
+// Scope: the retry happens inside one gateway request, so a stream is only
+// retried before its first chunk reaches the client (the kernel already holds
+// the first chunk back), and a non-stream call is retried as a whole.
+type RetryPolicy struct {
+	OnStatus       []int    `yaml:"on_status,omitempty" json:"on_status,omitempty"`
+	OnBodyContains []string `yaml:"on_body_contains,omitempty" json:"on_body_contains,omitempty"`
+	Delay          string   `yaml:"delay,omitempty" json:"delay,omitempty"`
+	MaxAttempts    int      `yaml:"max_attempts,omitempty" json:"max_attempts,omitempty"`
+}
+
+// ShouldRetry reports whether one failed attempt matches this policy.
+func (r *RetryPolicy) ShouldRetry(status int, body string) bool {
+	if r == nil || r.MaxAttempts <= 1 {
+		return false
+	}
+	statusOK := len(r.OnStatus) == 0
+	for _, s := range r.OnStatus {
+		if s == status {
+			statusOK = true
+			break
+		}
+	}
+	if !statusOK {
+		return false
+	}
+	if len(r.OnBodyContains) == 0 {
+		return true
+	}
+	for _, sub := range r.OnBodyContains {
+		if sub != "" && strings.Contains(body, sub) {
+			return true
+		}
+	}
+	return false
+}
+
+// DelayDuration is the configured pause between attempts (default 10s).
+func (r *RetryPolicy) DelayDuration() time.Duration {
+	if r == nil || r.Delay == "" {
+		return 10 * time.Second
+	}
+	d, err := time.ParseDuration(r.Delay)
+	if err != nil || d < 0 {
+		return 10 * time.Second
+	}
+	return d
+}
+
 // Load reads and validates a config file. When the file does not exist yet a
 // default config is generated at that path first (first-run bootstrap), so a
-// fresh binary just works: `llmsproxy -config /path/to/config.yaml`.
+// fresh binary just works: `modelrouter -config /path/to/config.yaml`.
 func Load(path string) (*Config, error) {
 	if _, err := EnsureDefault(path); err != nil {
 		return nil, err
@@ -423,182 +477,6 @@ type GWKey struct {
 // chain" so a user can clear an override back to the global default.
 func (k GWKey) HasOwnAuto() bool { return len(k.Auto) > 0 }
 
-// BillingDSL holds declarative per-URL pricing profiles for the billing plugin.
-//
-// Profiles are the "let the user pick" axis: the same upstream URL can be
-// declared under multiple profiles, and the operator chooses which is active.
-// Switching recomputes and re-injects the billing plugin's prices table.
-type BillingDSL struct {
-	// Active is the id of the profile in effect. Empty => the first profile.
-	Active   string           `yaml:"active,omitempty" json:"active,omitempty"`
-	Profiles []BillingProfile `yaml:"profiles,omitempty" json:"profiles,omitempty"`
-}
-
-// BillingProfile is one named set of per-URL pricing rules.
-type BillingProfile struct {
-	ID    string `yaml:"id" json:"id"`
-	Label string `yaml:"label,omitempty" json:"label,omitempty"`
-	// Default applies when no rule matches: "free" or "unpriced".
-	Default string        `yaml:"default,omitempty" json:"default,omitempty"`
-	Rules   []BillingRule `yaml:"rules,omitempty" json:"rules,omitempty"`
-}
-
-// BillingRule declares the pricing for everything under one URL.
-type BillingRule struct {
-	// URL matches a source's base_url. "*" = any unmatched URL.
-	URL string `yaml:"url" json:"url"`
-	// Mode is one of: free, token, subscription, unpriced.
-	Mode     string `yaml:"mode" json:"mode"`
-	Currency string `yaml:"currency,omitempty" json:"currency,omitempty"`
-	// Subscription is the flat monthly cost (for mode=subscription).
-	Subscription float64 `yaml:"subscription,omitempty" json:"subscription,omitempty"`
-	// Models declares per-token prices when mode=token.
-	Models map[string]BillingToken `yaml:"models,omitempty" json:"models,omitempty"`
-	// Free is true when mode=free (kept for explicitness).
-	// Peak applies to ALL token-priced requests under this URL.
-	Peak *BillingPeak `yaml:"peak,omitempty" json:"peak,omitempty"`
-}
-
-// BillingToken is the per-model price.
-type BillingToken struct {
-	Prompt        string   `yaml:"prompt" json:"prompt"`         // USD per million tokens, e.g. "0.15"
-	Completion    string   `yaml:"completion" json:"completion"` // USD per million tokens
-	CacheDiscount *float64 `yaml:"cache_discount,omitempty" json:"cache_discount,omitempty"`
-	// Reasoning is the USD-per-million rate for thinking tokens, for the rare
-	// provider that charges them differently from ordinary output. Omit it and
-	// reasoning is billed at Completion — which is what OpenAI, DeepSeek and
-	// Anthropic all do, and what every source on this deployment reports.
-	//
-	// It is a rate for a SUBSET of completion tokens, never an extra charge: the
-	// billing plugin applies only the difference between this and Completion, so
-	// setting it to a multiple of Completion reprices the thinking share rather
-	// than billing it twice.
-	Reasoning string `yaml:"reasoning,omitempty" json:"reasoning,omitempty"`
-	// LongContext is the prompt-price step above a token threshold. Nil = one
-	// flat rate, which is what a rule means when it says nothing.
-	LongContext *BillingLongContext `yaml:"long_context,omitempty" json:"long_context,omitempty"`
-}
-
-// BillingPeak declares peak-hour pricing.
-type BillingPeak struct {
-	Multiplier float64 `yaml:"multiplier" json:"multiplier"`
-	// Weekdays: 1=Mon..5=Fri (Lua os.date !%w, 0=Sun)
-	Weekdays []int `yaml:"weekdays,omitempty" json:"weekdays,omitempty"`
-	// Hours: each pair is [lo,hi] inclusive
-	Hours [][2]int `yaml:"hours,omitempty" json:"hours,omitempty"`
-}
-
-// BillingLongContext is a prompt-price step for long inputs.
-//
-// Every major provider charges the input tokens ABOVE a threshold at a higher
-// rate — DeepSeek doubles past 128k, OpenAI and Anthropic step up as well. A
-// flat prompt price therefore under-bills exactly the traffic an agent workload
-// generates, and it under-bills it silently: the number is plausible, the bill
-// adds up, and nothing says the rate was wrong.
-//
-// The step applies to the PROMPT only. Output and cache reads are billed at
-// their own rates regardless of input length, which is how providers publish
-// it — the higher rate is for the extra context you made the model read, not for
-// what it wrote back.
-type BillingLongContext struct {
-	// Above is the token count the step starts at. A request with exactly this
-	// many prompt tokens is still at the base rate; the step covers the tokens
-	// BEYOND it. Inclusive-below matches how the providers document it.
-	Above int `yaml:"above" json:"above"`
-	// Prompt is the USD-per-million rate for the tokens above Above.
-	Prompt string `yaml:"prompt" json:"prompt"`
-	// Multiplier is an alternative to Prompt: a factor applied to the base rate
-	// (DeepSeek's published step is exactly 2x). When both are set Prompt wins,
-	// and setting neither is a validation error rather than a silent no-op.
-	Multiplier float64 `yaml:"multiplier,omitempty" json:"multiplier,omitempty"`
-}
-
-// Validate rejects a BillingDSL that could not work as written.
-func (b *BillingDSL) Validate() error {
-	if b == nil {
-		return nil
-	}
-	ids := map[string]bool{}
-	for i := range b.Profiles {
-		p := &b.Profiles[i]
-		if p.ID == "" {
-			return fmt.Errorf("billing profile [%d] has no id", i)
-		}
-		if ids[p.ID] {
-			return fmt.Errorf("billing profile id %q appears more than once", p.ID)
-		}
-		ids[p.ID] = true
-		if p.Default != "" && p.Default != "free" && p.Default != "unpriced" {
-			return fmt.Errorf("profile %q: default must be free or unpriced, got %q", p.ID, p.Default)
-		}
-		for j := range p.Rules {
-			r := &p.Rules[j]
-			if r.URL == "" {
-				return fmt.Errorf("profile %q rule [%d]: url is required", p.ID, j)
-			}
-			switch r.Mode {
-			case "free", "token", "subscription", "unpriced":
-			default:
-				return fmt.Errorf("profile %q rule [%d]: mode must be free/token/subscription/unpriced, got %q", p.ID, j, r.Mode)
-			}
-			if r.Mode == "token" {
-				for m, t := range r.Models {
-					if t.Prompt == "" || t.Completion == "" {
-						return fmt.Errorf("profile %q model %q: prompt and completion are required for token mode", p.ID, m)
-					}
-					if _, err := ParseFloatUSDPerM(t.Prompt); err != nil {
-						return fmt.Errorf("profile %q model %q prompt: %w", p.ID, m, err)
-					}
-					if _, err := ParseFloatUSDPerM(t.Completion); err != nil {
-						return fmt.Errorf("profile %q model %q completion: %w", p.ID, m, err)
-					}
-				}
-			}
-		}
-	}
-	// Active must resolve to a profile id.
-	if b.Active != "" && !ids[b.Active] {
-		return fmt.Errorf("billing.active = %q but no profile has that id", b.Active)
-	}
-	return nil
-}
-
-// Resolve returns the profile with the given id; empty id (or an id that does
-// not exist) falls back to the FIRST profile. A config with profiles but no
-// active marker should mean "use the one I wrote first" rather than "price
-// nothing", which is what a nil here would silently do.
-func (b *BillingDSL) Resolve(id string) *BillingProfile {
-	if b == nil || len(b.Profiles) == 0 {
-		return nil
-	}
-	if id != "" {
-		for i := range b.Profiles {
-			if b.Profiles[i].ID == id {
-				return &b.Profiles[i]
-			}
-		}
-	}
-	return &b.Profiles[0]
-}
-
-// parseFloatUSDPerM parses a USD-per-million price string. Returns the
-// per-single-token rate the plugin expects (price / 1e6).
-//
-// Kept as a string in the DSL rather than a float: "0.15" survives round-trips
-// through YAML editors, and a malformed value must fail VALIDATION (with the
-// file and field named) rather than parse as 0 and under-bill silently — which
-// is exactly the failure mode that motivated the DSL.
-func ParseFloatUSDPerM(s string) (float64, error) {
-	f, err := strconv.ParseFloat(strings.TrimSpace(s), 64)
-	if err != nil {
-		return 0, fmt.Errorf("not a number: %q", s)
-	}
-	if f < 0 {
-		return 0, fmt.Errorf("negative price: %q", s)
-	}
-	return f / 1e6, nil
-}
-
 // KeyQuota is retained only to carry a scope entry's caps through the admin
 // API. Quotas are per model, never per key: there is deliberately no key-wide
 // total, so exhausting one model's budget never blocks the others.

internal/config/retry_policy_test.go

@@ -0,0 +1,59 @@
+package config
+
+import (
+	"testing"
+	"time"
+)
+
+// ShouldRetry 是纯数据判据：状态码列表 + 响应体子串。它决定"这次失败值不值得
+// 在网关内部等一会儿重试"。厂商知识绝不能进内核，所以判据必须能用一个表格测清楚。
+func TestRetryPolicyShouldRetry(t *testing.T) {
+	cases := []struct {
+		name   string
+		policy *RetryPolicy
+		status int
+		body   string
+		want   bool
+	}{
+		{"nil policy", nil, 500, "anything", false},
+		{"max_attempts=1 disables", &RetryPolicy{OnStatus: []int{400}, MaxAttempts: 1}, 400, "x", false},
+		{"status matches, no body filter",
+			&RetryPolicy{OnStatus: []int{429}, MaxAttempts: 3}, 429, "", true},
+		{"status mismatch",
+			&RetryPolicy{OnStatus: []int{429}, MaxAttempts: 3}, 400, "429", false},
+		{"status + body match",
+			&RetryPolicy{OnStatus: []int{400}, OnBodyContains: []string{"TM.00001041"}, MaxAttempts: 3},
+			400, `{"error_code":"TM.00001041"}`, true},
+		{"status matches but body does not",
+			&RetryPolicy{OnStatus: []int{400}, OnBodyContains: []string{"TM.00001041"}, MaxAttempts: 3},
+			400, `{"error_code":"TM.00001005"}`, false},
+		{"empty substring never matches",
+			&RetryPolicy{OnStatus: []int{400}, OnBodyContains: []string{""}, MaxAttempts: 3}, 400, "x", false},
+		{"any of several body signals",
+			&RetryPolicy{OnStatus: []int{400, 429}, OnBodyContains: []string{"TM.00001041", "TPM limit"}, MaxAttempts: 3},
+			429, `{"message":"TPM limit exceeded"}`, true},
+	}
+	for _, tc := range cases {
+		if got := tc.policy.ShouldRetry(tc.status, tc.body); got != tc.want {
+			t.Errorf("%s: ShouldRetry(%d, %q) = %v, want %v", tc.name, tc.status, tc.body, got, tc.want)
+		}
+	}
+}
+
+func TestRetryPolicyDelay(t *testing.T) {
+	cases := []struct {
+		name   string
+		policy *RetryPolicy
+		want   time.Duration
+	}{
+		{"nil", nil, 10 * time.Second},
+		{"explicit", &RetryPolicy{Delay: "250ms"}, 250 * time.Millisecond},
+		{"garbage falls back", &RetryPolicy{Delay: "not-a-duration"}, 10 * time.Second},
+		{"negative falls back", &RetryPolicy{Delay: "-5s"}, 10 * time.Second},
+	}
+	for _, tc := range cases {
+		if got := tc.policy.DelayDuration(); got != tc.want {
+			t.Errorf("%s: DelayDuration() = %v, want %v", tc.name, got, tc.want)
+		}
+	}
+}

internal/config/secret.go

@@ -27,8 +27,9 @@ func NewGatewayKey() (string, error) {
 
 // SecretBox encrypts and decrypts sensitive values (upstream API keys,
 // custom header values, gateway keys) for persist-time protection. The
-// master key comes from the LLMS_PROXY_MASTER_KEY environment variable
-// (hex, 64 chars) or from a generated <runtime-dir>/master.key file.
+// master key comes from the MODELROUTER_MASTER_KEY environment variable
+// (hex, 64 chars; the pre-rename LLMS_PROXY_MASTER_KEY is still accepted) or
+// from a generated <runtime-dir>/master.key file.
 type SecretBox struct {
 	aead cipher.AEAD
 }
@@ -37,10 +38,18 @@ type SecretBox struct {
 // file next to the runtime file. If neither exists, a fresh random key is
 // generated and written to master.key with mode 0600.
 func NewSecretBox(runtimeFile string) (*SecretBox, error) {
-	if v := os.Getenv("LLMS_PROXY_MASTER_KEY"); v != "" {
-		key, err := hex.DecodeString(strings.TrimSpace(v))
+	// New name first, then the pre-rename name: an existing unit file that still
+	// exports LLMS_PROXY_MASTER_KEY must keep starting after an upgrade.
+	envKey := os.Getenv("MODELROUTER_MASTER_KEY")
+	envName := "MODELROUTER_MASTER_KEY"
+	if envKey == "" {
+		envKey = os.Getenv("LLMS_PROXY_MASTER_KEY")
+		envName = "LLMS_PROXY_MASTER_KEY"
+	}
+	if envKey != "" {
+		key, err := hex.DecodeString(strings.TrimSpace(envKey))
 		if err != nil || len(key) != 32 {
-			return nil, fmt.Errorf("LLMS_PROXY_MASTER_KEY: expected 64 hex chars")
+			return nil, fmt.Errorf("%s: expected 64 hex chars", envName)
 		}
 		return newBox(key)
 	}

internal/core/channels_authorize_test.go

@@ -6,7 +6,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // callback 平台的授权 URL 是最容易"看起来对、实际不可用"的地方：URL 能拼出来、

internal/core/channels_callback_loopback_test.go

@@ -9,7 +9,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 三个 callback 平台的回调路径**各是注册值**，且校验规则不同：

internal/core/channels_clear_credentialless_test.go

@@ -3,7 +3,7 @@ package core
 import (
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ 无凭据账号的批量清理：**只删没有凭据的**，有凭据的一律不动。

internal/core/channels_codearts_catalog_test.go

@@ -8,7 +8,7 @@ import (
 	"sync"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ codearts 的模型目录必须从**两个**签名端点拉取并合并。

internal/core/channels_codearts_oauth_flow_test.go

@@ -8,7 +8,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // CodeArts 端到端：起真实回调服务器 → 模上游回 code → 走真实 STS 交换

internal/core/channels_codearts_port_test.go

@@ -4,7 +4,7 @@ import (
 	"fmt"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // CodeArts 的 portal 拒绝 <10000 的回调端口（参考 listenOnCallbackPort 的

internal/core/channels_lobster_claim_test.go

@@ -7,7 +7,7 @@ import (
 	"net/http/httptest"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 有道每日活动领取是**三步流程**，参考注释点名：无可用活动是**正常业务状态**，

internal/core/channels_lobster_exchange_test.go

@@ -2,7 +2,7 @@ package core
 
 import (
 	"encoding/json"
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 	"net/http"
 	"net/http/httptest"
 	"testing"

internal/core/channels_lobster_loopback_test.go

@@ -10,7 +10,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 有道登录闭环的判据：portal 会把授权重定向回 **本机 loopback 接收器**

internal/core/channels_loomy_test.go

@@ -8,7 +8,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // loomyHarness：Loomy 微信扫码长轮询的测试台。

internal/core/channels_models_remote_test.go

@@ -8,7 +8,7 @@ import (
 	"sync"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 远端模型目录的判据：**远端为主 + 兜底补位**（照参考 reconcileWithFallback）。

internal/core/channels_more_test.go

@@ -9,7 +9,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // channelsMoreTest：新引入的三条真实流程 + 两个边界流的行为对齐测试。

internal/core/channels_orphan_callback_test.go

@@ -5,12 +5,12 @@ import (
 	"path/filepath"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ 插件加载时必须清掉上一进程留下的孤儿回调文件。
 //
-// 真实事故：生产 `/etc/llmsproxy/plugins/` 攒了 **18 个 callback-*.port**
+// 真实事故：生产 `/etc/modelrouter/plugins/` 攒了 **18 个 callback-*.port**
 // （对应的 .json 早已消失）。它们永远不会被清，因为：
 //
 //	· loopbacks 追踪表只活在内存（plugin.state.loopbacks）；
@@ -25,7 +25,7 @@ func TestPluginLoadSweepsOrphanCallbackFiles(t *testing.T) {
 	if err := os.MkdirAll(pluginDir, 0755); err != nil {
 		t.Fatal(err)
 	}
-	t.Setenv("LLMSPROXY_PLUGIN_DIR", pluginDir)
+	t.Setenv("MODELROUTER_PLUGIN_DIR", pluginDir)
 
 	// 摆上一轮的残留（模拟上一进程被强杀后的现场）
 	orphans := []string{

internal/core/channels_platforms_logged_in_test.go

@@ -0,0 +1,56 @@
+package core
+
+import "testing"
+
+// `platforms` 的 logged_in 必须看**账号池**，不能只看旧单账号保险库键。
+//
+// 真实缺陷（2026-10-20 用户报「codearts 还是不可用」）：
+// `list_platforms` 当时写的是 `credentialOf("channels:" .. id)` —— 那是旧版
+// 单账号的键；而现网凭据键是 `channels:<platform>:<accountId>`（一平台多账号）。
+// 于是**所有**通过账号池接入的平台都被判成未登录，UI 把渠道卡片渲染成"未接入"，
+// 用户看到的正是"渠道不可用"，而底层模型其实完全能用。
+//
+// 判据钉两层：池里的账号让 logged_in=true，且账号标识能被 UI 拿到。
+func TestPlatformsLoggedInFromAccountPool(t *testing.T) {
+	_, ps := poolCore(t)
+	if _, err := ps.Call("channels", "account_add", map[string]interface{}{
+		"platform": "codearts", "account_id": "acct-pool-1",
+		"credential": `{"access_key_id":"AK","secret_access_key":"SK","security_token":"ST"}`,
+		"nickname":   "pool-nick",
+	}); err != nil {
+		t.Fatalf("account_add: %v", err)
+	}
+	// list_platforms 是 HTTP 路由处理器（不在 plugin.exports 里），走探针取同一份数据
+	m, err := ps.Call("channels", "__platforms_probe", map[string]interface{}{})
+	if err != nil {
+		t.Fatalf("__platforms_probe: %v", err)
+	}
+	w, _ := m.(map[string]interface{})
+	j, _ := w["json"].(map[string]interface{})
+	arr, _ := j["platforms"].([]interface{})
+	if len(arr) == 0 {
+		t.Fatalf("platforms 不应为空")
+	}
+	var got map[string]interface{}
+	for _, raw := range arr {
+		p, _ := raw.(map[string]interface{})
+		if p["id"] == "codearts" {
+			got = p
+		}
+	}
+	if got == nil {
+		t.Fatalf("platforms 里找不到 codearts")
+	}
+	if got["logged_in"] != true {
+		t.Fatalf("账号池里有凭据的账号，logged_in 必须为 true；实得 %#v", got)
+	}
+	if got["account_id"] != "acct-pool-1" {
+		t.Errorf("account_id 应回池里的账号 id；实得 %#v", got["account_id"])
+	}
+	if got["account"] != "pool-nick" {
+		t.Errorf("account 应回池里的昵称；实得 %#v", got["account"])
+	}
+	if s, _ := got["accounts"].(float64); s < 1 {
+		t.Errorf("accounts 计数应 >=1；实得 %#v", got["accounts"])
+	}
+}

internal/core/channels_plugin_test.go

@@ -4,7 +4,7 @@ import (
 	"encoding/json"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // TestChannelsPluginEndToEnd 是「同类插件」的完整验收：

internal/core/channels_pool_reason_test.go

@@ -9,7 +9,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ 「为什么没有可用账号」必须区分**真的用尽**与**查不到**。

internal/core/channels_pool_test.go

@@ -4,7 +4,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 
 	"fmt"
 	"net/http"

internal/core/channels_qoder_adapter_test.go

@@ -4,7 +4,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // qoderReason 调真实的 transform_error 钩子（不自己重算）。

internal/core/channels_qoder_claim_test.go

@@ -9,7 +9,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // Qoder 活动领取有三条参考用**抓包 + 用户报障**定位过的「会误报」陷阱，

internal/core/channels_qoder_test.go

@@ -8,7 +8,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // qoderHarness：Qoder PKCE 设备码轮的测试台。

internal/core/channels_raccoon_test.go

@@ -8,7 +8,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // raccoonHarness：Raccoon 内置二维码扫码登录的测试台。

internal/core/channels_remove_source_perm_test.go

@@ -6,7 +6,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ 删除账号必须**真的把源撤掉** —— 且撤不掉时不能静默。
@@ -190,8 +190,14 @@ func TestChannelsDeclaresEveryKernelActionItCalls(t *testing.T) {
 
 	// ② 抽出实际调用的动作
 	called := map[string]bool{}
-	for _, m := range regexp.MustCompile(`__llmsproxy_kernel\.call\(\s*"([a-z_]+)"`).FindAllStringSubmatch(s, -1) {
-		called[m[1]] = true
+	// 插件只允许走公开 SDK（modelrouter.kernel）；旧名 __llmsproxy_kernel 是
+	// 兼容别名，新代码用它就说明插件又在依赖内核私有全局了。
+	for _, m := range regexp.MustCompile(`\b(?:modelrouter|__llmsproxy_kernel)\.kernel\.call\(\s*"([a-z_]+)"|\b__llmsproxy_kernel\.call\(\s*"([a-z_]+)"`).FindAllStringSubmatch(s, -1) {
+		name := m[1]
+		if name == "" {
+			name = m[2] // 旧名的分支
+		}
+		called[name] = true
 	}
 	if len(called) == 0 {
 		t.Fatal("没扫描到任何内核调用（判据失效）")

internal/core/channels_reorder_effect_test.go

@@ -3,7 +3,7 @@ package core
 import (
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ account_reorder 必须**真的改变内核里的源顺序**。

internal/core/channels_store_atomic_test.go

@@ -33,8 +33,8 @@ func TestAccountPoolStoreIsAtomicAndConsistent(t *testing.T) {
 	storePath := filepath.Join(dir, ".channels-store.json")
 	t.Logf("插件存储目录: %s", dir)
 	// ⚠️ 必须在**测试目录**里，绝不能是生产路径
-	if _, err := os.Stat("/etc/llmsproxy/plugins/.channels-store.json"); err == nil {
-		if filepath.Clean(dir) == "/etc/llmsproxy/plugins" {
+	if _, err := os.Stat("/etc/modelrouter/plugins/.channels-store.json"); err == nil {
+		if filepath.Clean(dir) == "/etc/modelrouter/plugins" {
 			t.Fatal("测试正在使用生产存储目录 —— 隔离失效（这会造成真实数据事故）")
 		}
 	}

internal/core/channels_store_test.go

@@ -7,7 +7,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // TestChannelsStoreIsPluginOwned 是「插件自持持久化」的回归。
@@ -23,7 +23,7 @@ func TestChannelsStoreIsPluginOwned(t *testing.T) {
 	// 固定插件目录：两次「重启」必须共用同一目录，否则测不出持久化。
 	// 这也是防止测试写进生产插件目录的那道隔离（kernelTestCore 会 respect 它）。
 	dir := t.TempDir()
-	t.Setenv("LLMSPROXY_PLUGIN_DIR", dir)
+	t.Setenv("MODELROUTER_PLUGIN_DIR", dir)
 	load := func() *lua.Plugins {
 		c := kernelTestCore(t)
 		ps := c.Plugins()
@@ -128,6 +128,10 @@ func TestChannelsMigratesLegacyKernelState(t *testing.T) {
 
 	c := kernelTestCore(t)
 	dir = c.Config().PluginDir
+	// 网关不再自动创建 plugin_dir（插件不由内核安装），测试自己建。
+	if err := os.MkdirAll(dir, 0o755); err != nil {
+		t.Fatal(err)
+	}
 	legacyPath := filepath.Join(dir, ".channels.state.json")
 	if err := os.WriteFile(legacyPath, b, 0o600); err != nil {
 		t.Fatal(err)
@@ -179,7 +183,7 @@ func TestChannelsMigrationKeepsLegacyOnWriteFailure(t *testing.T) {
 	// （<dir>/.channels-store.json.tmp）**占成一个目录**，io.open(tmp,"w") 必失败。
 	// 不用 chmod：测试以 root 跑，root 绕过权限位，chmod 拦不住。
 	sub := t.TempDir()
-	t.Setenv("LLMSPROXY_PLUGIN_DIR", sub)
+	t.Setenv("MODELROUTER_PLUGIN_DIR", sub)
 
 	legacy := map[string]interface{}{
 		"version": 1,

internal/core/channels_stress_test.go

@@ -11,8 +11,8 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/config"
+	"modelrouter/internal/lua"
 )
 
 // mustBox 构造一个真 SecretBox（与运行期同一实现，密钥落 runtime dir）。

internal/core/channels_trae_catalog_test.go

@@ -9,7 +9,7 @@ import (
 	"sync"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ TRAE 的 batch 模型目录解析：两个**真实缺陷**必须在移植时一起带上
@@ -120,23 +120,23 @@ func newTraeCatalogHarness(t *testing.T, onBody func(string)) *traeCatalogHarnes
 			// ① solo_agent：glm-5.2 有档位；另有三个必被硬性过滤的条目
 			`{"function":"solo_agent","config_info_list":[`+
 			`{"config_name":"glm-5.2","usage":"chat_completion","config_switch":true,`+
-			`"reasoning_effort_config":{"support_thinking":true,"options":["high","low"]}},`+
+			`"reasoning_effort_config":{"support_thinking":true,"options":["high","low"]}}`+
+			`]},`+
+			// ② solo_work_lite 排在**后面**：glm-5.2 在这里**没有**档位配置。
+			//    另三个"必被过滤"的条目也放这里 —— 生产里适配器恒发这条通道，
+			//    硬性过滤的判据必须在这条通道上被验证到（放 solo_agent 会因为
+			//    admit_functions 策略先行剔除，测试就测不到过滤逻辑了）。
+			`{"function":"solo_work_lite","config_info_list":[`+
+			`{"config_name":"glm-5.2","usage":"chat_completion","config_switch":true},`+
 			`{"config_name":"summary-model","usage":"summary","config_switch":true},`+
 			`{"config_name":"disabled-model","usage":"chat_completion","config_switch":false},`+
 			`{"config_name":"hidden-model","usage":"chat_completion","config_switch":true,"is_invisible_to_user":true},`+
 			`{"config_name":"no-function-model","usage":"chat_completion","config_switch":true},`+
-			// is_custom_model=true ⇒ 上游以流内 4001 拒绝（"仅可见但不可调用"）
 			`{"config_name":"custom-flag-model","usage":"chat_completion","config_switch":true,`+
 			`"display_config":{"is_custom_model":true}},`+
-			// 对照组：名字带 custom_model_ 前缀但**标志为 false** ⇒ 必须**保留**
-			// （参考点名：判据是标志的值，不是模型名）
 			`{"config_name":"custom_model_prefix_but_ok","usage":"chat_completion","config_switch":true,`+
 			`"display_config":{"is_custom_model":false}}`+
 			`]},`+
-			// ② solo_work_lite 排在**后面**：glm-5.2 在这里**没有**档位配置
-			`{"function":"solo_work_lite","config_info_list":[`+
-			`{"config_name":"glm-5.2","usage":"chat_completion","config_switch":true}`+
-			`]},`+
 			// ③ 白名单外通道：整组必须被丢弃
 			`{"function":"inline_chat","config_info_list":[`+
 			`{"config_name":"only-in-bad-channel","usage":"chat_completion","config_switch":true}`+

internal/core/channels_trae_chat_headers_test.go

@@ -4,7 +4,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ TRAE 的**推理请求**必须用 Cloud-IDE-JWT 头族，不是 Bearer。

internal/core/channels_trae_checkin_test.go

@@ -6,7 +6,7 @@ import (
 	"net/http/httptest"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // Trae 签到是最容易被"看起来对"骗过的地方：请求能发出去、上游可能也返回 200，

internal/core/channels_trae_derive_test.go

@@ -6,7 +6,7 @@ import (
 	"os"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 手工移植的确定性派生必须与参考**逐字节一致**，否则签名的设备身份是错的

internal/core/channels_trae_exchange_shape_test.go

@@ -6,7 +6,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ TRAE ExchangeToken 的凭据在 **Result 里**，不在顶层。

internal/core/channels_trae_pkce_test.go

@@ -9,7 +9,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ TRAE 回调只带 code（PKCE 形态）时必须**明确报出**「暂不支持 PKCE」，

internal/core/channels_trae_test.go

@@ -10,7 +10,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // traeHarness：TRAE callback 流的测试台。假上游模拟 ExchangeToken；

internal/core/channels_vnc_test.go

@@ -5,7 +5,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // TestChannelsVncStackEndToEnd 验证 VNC 兜底链路在**真插件**上跑通：

internal/core/channels_xdisplay_test.go

@@ -6,7 +6,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // 显示号泄漏是「测试单独跑绿、全量跑红」的根源，值得一条判据钉住它。
@@ -34,7 +34,7 @@ func TestXDisplaySelectionReclaimsStaleLocks(t *testing.T) {
 	// （见 kernelTestCore）；硬编码 90 段会让本判据造出**根本不会被扫描到**的
 	// 陈旧锁，于是"没复用"这个断言永远失败 —— 那是判据与被测对象脱节。
 	base := 300
-	if v := os.Getenv("LLMSPROXY_XDISPLAY_BASE"); v != "" {
+	if v := os.Getenv("MODELROUTER_XDISPLAY_BASE"); v != "" {
 		if n, err := strconv.Atoi(v); err == nil && n > 0 {
 			base = n
 		}

internal/core/channels_zcode_identity_test.go

@@ -7,7 +7,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // ★ ZCode 的身份块是 3012 准入的**唯一开关**。
@@ -269,7 +269,7 @@ func TestZcodeAdapterFailsLoudlyWithoutAnthropicBase(t *testing.T) {
 	if err := os.MkdirAll(pluginDir, 0755); err != nil {
 		t.Fatal(err)
 	}
-	t.Setenv("LLMSPROXY_PLUGIN_DIR", pluginDir)
+	t.Setenv("MODELROUTER_PLUGIN_DIR", pluginDir)
 
 	c := kernelTestCore(t)
 	ps := c.Plugins()

internal/core/channels_zcode_test.go

@@ -10,7 +10,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // zcodeHarness：ZCode 两段式服务端中介授权流的测试台。

internal/core/core.go

@@ -16,11 +16,10 @@ import (
 	"sync/atomic"
 	"time"
 
-	"llmsproxy/internal/billing"
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/lua"
-	"llmsproxy/internal/provider"
-	"llmsproxy/internal/scheduler"
+	"modelrouter/internal/config"
+	"modelrouter/internal/lua"
+	"modelrouter/internal/provider"
+	"modelrouter/internal/scheduler"
 )
 
 // Core owns the running configuration and adapters. mu guards every read and
@@ -96,9 +95,6 @@ func NewFromConfig(cfg *config.Config) (*Core, error) {
 	for name, aud := range cfg.PluginVisibility {
 		c.plugins.SetPluginVisibility(name, aud)
 	}
-	if err := c.plugins.SeedBundled(); err != nil {
-		log.Printf("[core] seed bundled plugins: %v", err)
-	}
 	if err := c.plugins.LoadDir(); err != nil {
 		log.Printf("[core] plugin dir: %v", err)
 	}
@@ -110,7 +106,6 @@ func NewFromConfig(cfg *config.Config) (*Core, error) {
 		log.Printf("[core] plugin %v %v loaded (hooks=%v ui=%v)",
 			p["name"], p["version"], p["hooks"], p["ui"])
 	}
-	c.applyBillingDSL()
 	c.store = config.NewStore(cfg.RuntimeFile)
 	// Share one box between the runtime store and config.yaml so a single
 	// master.key seals both files. config.Load left the config holding
@@ -1010,63 +1005,6 @@ func (c *Core) Close() {
 	}
 }
 
-// applyBillingDSL compiles the configured billing profile and injects it into
-// the billing plugin.
-//
-// Configured pricing replaces the hand-written JSON blob that used to be PUT
-// through the state API. That path was used twice in production and both times
-// it "succeeded" while being wrong: once without the required `prices` wrapper
-// (silently wiping the accumulated totals), once with a peak-window shape the
-// plugin does not read (peak traffic billed at off-peak rates, no error). A
-// typed config with load-time validation turns both mistakes into startup
-// errors naming the file and field.
-//
-// A DSL with no profile, or one that fails to compile, leaves the plugin's
-// prices untouched: billing keeps running on whatever it had, and the problem
-// is logged rather than becoming a startup failure — the gateway must forward
-// even when its cost accounting is misconfigured.
-func (c *Core) applyBillingDSL() {
-	ps := c.plugins
-	if ps == nil || ps.Count() == 0 {
-		return
-	}
-	// The billing plugin may simply not be installed; a DSL without it is a
-	// no-op, not an error.
-	var have bool
-	for _, row := range ps.List() {
-		if row["name"] == "billing" && row["loaded"] == true {
-			have = true
-			break
-		}
-	}
-	if !have {
-		return
-	}
-	dsl := c.cfg.BillingDSL
-	if dsl == nil || len(dsl.Profiles) == 0 {
-		return
-	}
-	if err := dsl.Validate(); err != nil {
-		log.Printf("[billing] config invalid, profile not applied: %v", err)
-		return
-	}
-	profile := dsl.Resolve(dsl.Active)
-	if profile == nil {
-		log.Printf("[billing] no billing profile resolved (active=%q)", dsl.Active)
-		return
-	}
-	prices, err := billing.Compile(profile, c.cfg.Sources)
-	if err != nil {
-		log.Printf("[billing] profile %q not applied: %v", profile.ID, err)
-		return
-	}
-	if err := ps.SetState("billing", map[string]interface{}{"prices": prices}); err != nil {
-		log.Printf("[billing] applying profile %q: %v", profile.ID, err)
-		return
-	}
-	log.Printf("[billing] profile %q applied (%d rules, %d sources)", profile.ID, len(profile.Rules), len(c.cfg.Sources))
-}
-
 // ValidateScopeQuotas checks every scope entry's caps before they are stored.
 // A typo in a period must be rejected at write time rather than silently
 // becoming a never-resetting budget — the opposite of what was typed.

internal/core/core_test.go

@@ -5,7 +5,7 @@ import (
 	"sync"
 	"testing"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // newTestConfig returns an in-memory config backed by temp files so

internal/core/device_harness_test.go

@@ -8,7 +8,7 @@ import (
 	"sync/atomic"
 	"testing"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // readBundledChannels 读内置 channels 插件源（供行为/变异测试共用）。

internal/core/kernel_actions.go

@@ -6,8 +6,8 @@ import (
 	"fmt"
 	"strings"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/config"
+	"modelrouter/internal/lua"
 )
 
 // 内核动作实现（R4-c）

internal/core/kernel_actions_test.go

@@ -6,8 +6,8 @@ import (
 	"path/filepath"
 	"testing"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/config"
+	"modelrouter/internal/lua"
 )
 
 // 本文件验证 R4-c/R4-d 的闭环：**插件自己驱动外部源的链接认证，然后把源接进内核**。
@@ -22,12 +22,12 @@ func kernelTestCore(t *testing.T) *Core {
 	dir := t.TempDir()
 	pluginDir := filepath.Join(dir, "plugins")
 	// 插件的自有存储位置由插件自己决定（读 config.yaml 的 plugin_dir，或
-	// LLMSPROXY_PLUGIN_DIR 覆盖）。测试**必须**指向临时目录 —— 否则插件会读到
-	// /etc/llmsproxy/config.yaml 并写进**生产插件目录**（真实事故：一次全量测试
+	// MODELROUTER_PLUGIN_DIR 覆盖）。测试**必须**指向临时目录 —— 否则插件会读到
+	// /etc/modelrouter/config.yaml 并写进**生产插件目录**（真实事故：一次全量测试
 	// 把 26 个测试账号写进了生产 .channels-store.json）。
 	// 调用方已显式设定时尊重它：多次「重启」要共用同一目录才能测持久化。
-	if os.Getenv("LLMSPROXY_PLUGIN_DIR") == "" {
-		t.Setenv("LLMSPROXY_PLUGIN_DIR", pluginDir)
+	if os.Getenv("MODELROUTER_PLUGIN_DIR") == "" {
+		t.Setenv("MODELROUTER_PLUGIN_DIR", pluginDir)
 	}
 	// ★ 显示器号与端口用**高段号**，避开生产。
 	//
@@ -36,8 +36,8 @@ func kernelTestCore(t *testing.T) *Core {
 	// 并去抢，结果 `Xvfb 未就绪（/tmp/.X11-unix/X91 未出现）`，而生产上正跑着
 	// 一套浏览器栈。判据本身没错，错的是**两个 /tmp 互不可见**而号段重叠。
 	// 故测试整体挪到 300+。
-	if os.Getenv("LLMSPROXY_XDISPLAY_BASE") == "" {
-		t.Setenv("LLMSPROXY_XDISPLAY_BASE", "300")
+	if os.Getenv("MODELROUTER_XDISPLAY_BASE") == "" {
+		t.Setenv("MODELROUTER_XDISPLAY_BASE", "300")
 	}
 	cfg := &config.Config{
 		Path:         filepath.Join(dir, "config.yaml"),

internal/core/kernel_no_deadlock_test.go

@@ -4,7 +4,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/lua"
+	"modelrouter/internal/lua"
 )
 
 // TestNoKernelActionDeadlocksFromLua 对**每个**内核动作做同一条断言：

internal/core/key_auto_concurrency_test.go

@@ -7,7 +7,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // AutoChainFor sits on the request hot path and reads keyAutoChains with no

internal/core/key_auto_test.go

@@ -4,8 +4,8 @@ import (
 	"path/filepath"
 	"testing"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/scheduler"
+	"modelrouter/internal/config"
+	"modelrouter/internal/scheduler"
 )
 
 // Per-key AUTO chains let an admin give one user their own scheduling chain

internal/core/source_hosts_test.go

@@ -4,7 +4,7 @@ import (
 	"net/url"
 	"testing"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // The plugin outbound allowlist is built from the hosts of configured sources.

internal/core/startup_secrets_test.go

@@ -6,7 +6,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // Startup must be idempotent, and every credential consumer must see PLAINTEXT.

internal/gateway/api.go

@@ -11,7 +11,7 @@ import (
 	"strings"
 	"time"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 type adapterPayload struct {
@@ -488,7 +488,7 @@ func (g *Gateway) handleStatsAPI(w http.ResponseWriter, r *http.Request) {
 		if to == 0 {
 			to = time.Now().UnixMilli()
 		}
-		csvHeaders(w, "llmsproxy-requests.csv")
+		csvHeaders(w, "modelrouter-requests.csv")
 		cw := csv.NewWriter(w)
 		names := map[string]string{}
 		for _, k := range g.core.ListKeys() {
@@ -537,7 +537,7 @@ func (g *Gateway) handleStatsAPI(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 	if r.URL.Query().Get("export") == "keys-csv" {
-		csvHeaders(w, "llmsproxy-keys.csv")
+		csvHeaders(w, "modelrouter-keys.csv")
 		cw := csv.NewWriter(w)
 		_ = cw.Write([]string{"key", "key_name", "role", "models", "total_requests", "success_requests", "failed_requests", "prompt_tokens", "completion_tokens", "total_tokens", "avg_latency_ms", "max_latency_ms", "created_at"})
 		keyFilter := exportKey(r)

internal/gateway/apiv1.go

@@ -31,7 +31,7 @@ import (
 	"sort"
 	"strings"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // apiV1Routes dispatches /api/v1/*. Kept separate from routes() so the original
@@ -144,7 +144,7 @@ func (g *Gateway) apiV1Index(w http.ResponseWriter, r *http.Request) {
 	}
 	index := map[string]interface{}{
 		"api_version": "v1",
-		"description": "llmsproxy management API. Authenticate with a gateway key: " +
+		"description": "modelrouter management API. Authenticate with a gateway key: " +
 			"'Authorization: Bearer <key>', '?api_key=<key>', or the gw_key cookie. " +
 			"Read endpoints accept any role; mutations require an admin key.",
 		"discovery": []ep{

internal/gateway/apiv1_test.go

@@ -7,7 +7,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // v1Gateway builds a gateway with one source and two keys, and returns the

internal/gateway/billing_rules_api.go

@@ -1,282 +0,0 @@
-package gateway
-
-import (
-	"encoding/json"
-	"net/http"
-
-	"llmsproxy/internal/billing"
-	"llmsproxy/internal/config"
-)
-
-// Billing rules API.
-//
-//	GET  /api/plugins/billing/rules    the active profile's rules + all profiles
-//	PUT  /api/plugins/billing/rules    replace the whole DSL and re-inject prices
-//	POST /api/plugins/billing/rules    { profile, rule } append one rule
-//	DELETE /api/plugins/billing/rules  { profile, url } remove one rule
-//
-// Why a separate endpoint instead of the generic plugin state: prices are
-// plugin state, but they are also a durable, reviewable CONFIGURATION. Routing
-// them through /state would let an admin key PUT arbitrary plugin state and
-// silently reset the accumulated accounting. Here the server owns the shape:
-// it validates the DSL, writes it back to config.yaml, recompiles, and hands
-// the plugin its prices table. The operator edits rules; the plugin keeps
-// numbers. The two are never mixed in one payload.
-//
-// Editing means "what the operator typed is what config.yaml holds". A
-// billing mistake found next week must be traceable to a reviewable file, not
-// to a blob in the plugin's state sidecar.
-func (g *Gateway) handleBillingRules(w http.ResponseWriter, r *http.Request) {
-	if reqRole(r.Context()) != "admin" {
-		writeError(w, http.StatusForbidden, "forbidden", "admin role required")
-		return
-	}
-	switch r.Method {
-	case http.MethodGet:
-		g.getBillingRules(w)
-	case http.MethodPut:
-		g.putBillingRules(w, r)
-	case http.MethodPost:
-		g.addBillingRule(w, r)
-	case http.MethodDelete:
-		g.delBillingRule(w, r)
-	default:
-		writeError(w, http.StatusMethodNotAllowed, "method_not_allowed", "use GET/PUT/POST/DELETE")
-	}
-}
-
-// billingRulesPayload is the UI-facing view: the profiles as declared, plus
-// which one is active and the source URLs they can match — the editor needs
-// the URL list to offer suggestions, and an operator typing a URL that matches
-// no source is exactly the silent-no-pricing failure this feature exists to
-// remove.
-type billingRulesPayload struct {
-	DSL      *config.BillingDSL `json:"billing"`
-	Active   string             `json:"active"`
-	URLs     []string           `json:"urls"`
-	Warnings []string           `json:"warnings"`
-}
-
-func (g *Gateway) billingRulesPayload() billingRulesPayload {
-	out := billingRulesPayload{}
-	cfg := g.core.Config()
-	if cfg != nil && cfg.BillingDSL != nil {
-		// Copy so the response cannot be mutated back through the pointer.
-		cp := *cfg.BillingDSL
-		out.DSL = &cp
-		if p := cp.Resolve(""); p != nil {
-			out.Active = p.ID
-		}
-	}
-	for _, s := range g.sourceURLs() {
-		out.URLs = append(out.URLs, s)
-	}
-	out.Warnings = g.billingRuleWarnings()
-	return out
-}
-
-func (g *Gateway) sourceURLs() []string {
-	cfg := g.core.Config()
-	if cfg == nil {
-		return nil
-	}
-	seen := map[string]bool{}
-	var out []string
-	for _, s := range cfg.Sources {
-		if s.BaseURL == "" || seen[s.BaseURL] {
-			continue
-		}
-		seen[s.BaseURL] = true
-		out = append(out, s.BaseURL)
-	}
-	return out
-}
-
-// billingRuleWarnings reports rules that match no configured source. This is
-// the silent killer of hand-written price tables: the rule parses, validates,
-// and prices nothing at all, so every request on that URL lands in unpriced.
-// Saying so out loud is the difference between a five-second fix and an
-// afternoon wondering why the bill is zero.
-func (g *Gateway) billingRuleWarnings() []string {
-	cfg := g.core.Config()
-	if cfg == nil || cfg.BillingDSL == nil {
-		return nil
-	}
-	urls := map[string]bool{}
-	for _, u := range g.sourceURLs() {
-		urls[u] = true
-	}
-	var warns []string
-	for _, p := range cfg.BillingDSL.Profiles {
-		for _, r := range p.Rules {
-			if r.URL == "*" || urls[r.URL] {
-				continue
-			}
-			warns = append(warns, "profile "+p.ID+": rule url "+r.URL+" matches no configured source")
-		}
-	}
-	return warns
-}
-
-func (g *Gateway) getBillingRules(w http.ResponseWriter) {
-	writeJSON(w, http.StatusOK, g.billingRulesPayload())
-}
-
-func (g *Gateway) putBillingRules(w http.ResponseWriter, r *http.Request) {
-	var body struct {
-		Billing *config.BillingDSL `json:"billing"`
-	}
-	if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
-		writeError(w, http.StatusBadRequest, "invalid_request", "invalid json: "+err.Error())
-		return
-	}
-	if body.Billing == nil {
-		writeError(w, http.StatusBadRequest, "invalid_request", "billing is required")
-		return
-	}
-	g.applyBillingDSLUpdate(w, body.Billing)
-}
-
-func (g *Gateway) addBillingRule(w http.ResponseWriter, r *http.Request) {
-	var body struct {
-		Profile string             `json:"profile"`
-		Rule    config.BillingRule `json:"rule"`
-	}
-	if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
-		writeError(w, http.StatusBadRequest, "invalid_request", "invalid json: "+err.Error())
-		return
-	}
-	cfg := g.core.Config()
-	if cfg == nil || cfg.BillingDSL == nil || len(cfg.BillingDSL.Profiles) == 0 {
-		writeError(w, http.StatusBadRequest, "no_profile", "no billing profiles configured; create one first")
-		return
-	}
-	next := cloneBillingDSL(cfg.BillingDSL)
-	idx := profileIndex(next, body.Profile)
-	if idx < 0 {
-		writeError(w, http.StatusBadRequest, "unknown_profile", "no such profile: "+body.Profile)
-		return
-	}
-	p := &next.Profiles[idx]
-	// Replace rather than append when the URL is already declared: two rules
-	// for one URL would silently make the first unreachable (first match wins),
-	// which is the exact ambiguity an editor should not be able to create.
-	replaced := false
-	for i := range p.Rules {
-		if p.Rules[i].URL == body.Rule.URL {
-			p.Rules[i] = body.Rule
-			replaced = true
-			break
-		}
-	}
-	if !replaced {
-		p.Rules = append(p.Rules, body.Rule)
-	}
-	g.applyBillingDSLUpdate(w, next)
-}
-
-func (g *Gateway) delBillingRule(w http.ResponseWriter, r *http.Request) {
-	var body struct {
-		Profile string `json:"profile"`
-		URL     string `json:"url"`
-	}
-	if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
-		writeError(w, http.StatusBadRequest, "invalid_request", "invalid json: "+err.Error())
-		return
-	}
-	cfg := g.core.Config()
-	if cfg == nil || cfg.BillingDSL == nil {
-		writeError(w, http.StatusBadRequest, "no_profile", "no billing profiles configured")
-		return
-	}
-	next := cloneBillingDSL(cfg.BillingDSL)
-	idx := profileIndex(next, body.Profile)
-	if idx < 0 {
-		writeError(w, http.StatusBadRequest, "unknown_profile", "no such profile: "+body.Profile)
-		return
-	}
-	p := &next.Profiles[idx]
-	out := p.Rules[:0]
-	found := false
-	for _, rule := range p.Rules {
-		if rule.URL == body.URL {
-			found = true
-			continue
-		}
-		out = append(out, rule)
-	}
-	if !found {
-		writeError(w, http.StatusNotFound, "not_found", "no rule for url "+body.URL)
-		return
-	}
-	p.Rules = out
-	g.applyBillingDSLUpdate(w, next)
-}
-
-// applyBillingDSLUpdate is the single write path: validate, persist to
-// config.yaml, recompile, inject, and only then report success. If the config
-// cannot be written the change is NOT applied in memory either — a rules editor
-// that says "saved" and loses the edit on the next restart is worse than one
-// that refuses.
-func (g *Gateway) applyBillingDSLUpdate(w http.ResponseWriter, next *config.BillingDSL) {
-	if err := next.Validate(); err != nil {
-		writeError(w, http.StatusBadRequest, "invalid_rules", err.Error())
-		return
-	}
-	cfg := g.core.Config()
-	if cfg == nil {
-		writeError(w, http.StatusInternalServerError, "no_config", "no config loaded")
-		return
-	}
-	// Assign BEFORE saving. The previous order saved first and assigned after,
-	// so every rule edit reported success while writing a config.yaml with no
-	// billing section at all — the operator's edit vanished on restart and
-	// nothing in the API response said so.
-	prev := cfg.BillingDSL
-	cfg.BillingDSL = next
-
-	// Compile before persisting: a profile that cannot be turned into prices
-	// must not reach the config file, or the next restart fails to load the
-	// very rules the editor just accepted.
-	prices, err := billing.CompileOpts(next.Resolve(next.Active), cfg.Sources, true)
-	if err != nil {
-		cfg.BillingDSL = prev // no half-applied state
-		writeError(w, http.StatusBadRequest, "compile_failed", err.Error())
-		return
-	}
-	if err := cfg.Save(); err != nil {
-		cfg.BillingDSL = prev
-		writeError(w, http.StatusInternalServerError, "persist_failed", err.Error())
-		return
-	}
-	ps := g.core.Plugins()
-	if ps != nil {
-		if err := ps.SetState("billing", map[string]interface{}{"prices": prices}); err != nil {
-			writeError(w, http.StatusBadRequest, "plugin_error", err.Error())
-			return
-		}
-	}
-	p := g.billingRulesPayload()
-	writeJSON(w, http.StatusOK, map[string]interface{}{
-		"ok": true, "billing": p.DSL, "active": p.Active, "warnings": p.Warnings,
-	})
-}
-
-func cloneBillingDSL(d *config.BillingDSL) *config.BillingDSL {
-	out := &config.BillingDSL{Active: d.Active}
-	out.Profiles = make([]config.BillingProfile, len(d.Profiles))
-	copy(out.Profiles, d.Profiles)
-	return out
-}
-
-func profileIndex(d *config.BillingDSL, id string) int {
-	if id == "" {
-		return -1
-	}
-	for i, p := range d.Profiles {
-		if p.ID == id {
-			return i
-		}
-	}
-	return -1
-}

internal/gateway/billing_rules_api_test.go

@@ -1,281 +0,0 @@
-package gateway
-
-import (
-	"encoding/json"
-	"net/http"
-	"net/http/httptest"
-	"os"
-	"sort"
-	"strings"
-	"testing"
-
-	"llmsproxy/internal/config"
-)
-
-// rulesGateway needs a real billing plugin (the API injects prices into it)
-// AND two sources whose base_urls the rules will match, so it cannot reuse
-// either existing helper: gatewayWithBilling has no sources, and
-// newTestGateway loads no plugins.
-func rulesGateway(t *testing.T) *Gateway {
-	t.Helper()
-	g := gatewayWithBilling(t)
-	cfg := g.core.Config()
-	cfg.Sources = []config.Source{
-		{Name: "localzen", BaseURL: "https://free.example.com/v1"},
-		{Name: "commandcode", BaseURL: "https://api.commandcode.ai/v1"},
-	}
-	return g
-}
-
-func rulesDo(t *testing.T, g *Gateway, method, body string) *httptest.ResponseRecorder {
-	t.Helper()
-	return doReq(t, g, method, "/api/plugins/billing/rules", body)
-}
-
-// TestBillingRulesCRUD covers the whole editor loop against the real HTTP
-// surface: read, add, replace-by-URL, delete. The replace-by-URL case is the
-// one worth pinning — two rules for one URL would make the first unreachable
-// (first match wins) without any error, which is exactly the kind of silent
-// ambiguity an editor must not be able to create.
-func TestBillingRulesCRUD(t *testing.T) {
-	g := rulesGateway(t)
-	seed := `{"billing":{"active":"p1","profiles":[{"id":"p1","rules":[
-      {"url":"https://free.example.com/v1","mode":"free"}]}]}}`
-	if rr := rulesDo(t, g, http.MethodPut, seed); rr.Code != http.StatusOK {
-		t.Fatalf("seed rules = %d: %s", rr.Code, rr.Body.String())
-	}
-
-	// Add a token-priced rule for the second URL.
-	add := `{"profile":"p1","rule":{"url":"https://api.commandcode.ai/v1","mode":"token",
-      "models":{"deepseek-v4.1-flash":{"prompt":"0.15","completion":"0.60"}}}}`
-	if rr := rulesDo(t, g, http.MethodPost, add); rr.Code != http.StatusOK {
-		t.Fatalf("add rule = %d: %s", rr.Code, rr.Body.String())
-	}
-	var got billingRulesPayload
-	if err := json.Unmarshal(rulesDo(t, g, http.MethodGet, "").Body.Bytes(), &got); err != nil {
-		t.Fatal(err)
-	}
-	if len(got.DSL.Profiles[0].Rules) != 2 {
-		t.Fatalf("after add there are %d rules, want 2", len(got.DSL.Profiles[0].Rules))
-	}
-
-	// Adding the SAME url again must replace, not append.
-	if rr := rulesDo(t, g, http.MethodPost, `{"profile":"p1","rule":{"url":"https://api.commandcode.ai/v1","mode":"free"}}`); rr.Code != http.StatusOK {
-		t.Fatalf("re-add same url = %d: %s", rr.Code, rr.Body.String())
-	}
-	if err := json.Unmarshal(rulesDo(t, g, http.MethodGet, "").Body.Bytes(), &got); err != nil {
-		t.Fatal(err)
-	}
-	if n := len(got.DSL.Profiles[0].Rules); n != 2 {
-		t.Fatalf("re-adding an existing url appended instead of replacing: %d rules", n)
-	}
-	var mode string
-	for _, r := range got.DSL.Profiles[0].Rules {
-		if r.URL == "https://api.commandcode.ai/v1" {
-			mode = r.Mode
-		}
-	}
-	if mode != "free" {
-		t.Errorf("re-added rule mode = %q, want free (the replacement must win)", mode)
-	}
-
-	// Delete by URL.
-	if rr := rulesDo(t, g, http.MethodDelete, `{"profile":"p1","url":"https://api.commandcode.ai/v1"}`); rr.Code != http.StatusOK {
-		t.Fatalf("delete rule = %d: %s", rr.Code, rr.Body.String())
-	}
-	if err := json.Unmarshal(rulesDo(t, g, http.MethodGet, "").Body.Bytes(), &got); err != nil {
-		t.Fatal(err)
-	}
-	if len(got.DSL.Profiles[0].Rules) != 1 {
-		t.Errorf("after delete there are %d rules, want 1", len(got.DSL.Profiles[0].Rules))
-	}
-}
-
-// TestBillingRulesRejectUnknownURLWithWarning is the silent-failure guard: a
-// rule whose URL matches no source prices nothing, so every request on it
-// falls into unpriced. The API must say so instead of accepting it silently.
-func TestBillingRulesRejectUnknownURLWithWarning(t *testing.T) {
-	g := rulesGateway(t)
-	if rr := rulesDo(t, g, http.MethodPut, `{"billing":{"active":"p","profiles":[{"id":"p",
-      "rules":[{"url":"https://typo.example.com/v1","mode":"free"}]}]}}`); rr.Code != http.StatusOK {
-		t.Fatalf("seed = %d: %s", rr.Code, rr.Body.String())
-	}
-	var got billingRulesPayload
-	if err := json.Unmarshal(rulesDo(t, g, http.MethodGet, "").Body.Bytes(), &got); err != nil {
-		t.Fatal(err)
-	}
-	if len(got.Warnings) == 0 {
-		t.Fatal("a rule matching no configured source was accepted with no warning — " +
-			"it will price nothing and every request will be recorded as unpriced")
-	}
-	if !strings.Contains(got.Warnings[0], "typo.example.com") {
-		t.Errorf("warning does not name the offending url: %v", got.Warnings)
-	}
-	// The editor needs the real URL list to offer suggestions.
-	if len(got.URLs) != 2 {
-		t.Errorf("urls offered to the editor = %v, want the 2 configured base_urls", got.URLs)
-	}
-}
-
-// TestBillingRulesInvalidIsRejectedNotPersisted checks the write path refuses
-// bad input and leaves the previous rules in place — an editor that clears the
-// table on a typo loses the price table.
-func TestBillingRulesInvalidIsRejectedNotPersisted(t *testing.T) {
-	g := rulesGateway(t)
-	seed := `{"billing":{"active":"p","profiles":[{"id":"p","rules":[
-      {"url":"https://free.example.com/v1","mode":"free"}]}]}}`
-	if rr := rulesDo(t, g, http.MethodPut, seed); rr.Code != http.StatusOK {
-		t.Fatalf("seed = %d: %s", rr.Code, rr.Body.String())
-	}
-	// An unknown mode must be refused.
-	if rr := rulesDo(t, g, http.MethodPost, `{"profile":"p","rule":{"url":"https://x/v1","mode":"banana"}}`); rr.Code != http.StatusBadRequest {
-		t.Errorf("invalid mode accepted with %d, want 400", rr.Code)
-	}
-	var got billingRulesPayload
-	if err := json.Unmarshal(rulesDo(t, g, http.MethodGet, "").Body.Bytes(), &got); err != nil {
-		t.Fatal(err)
-	}
-	if len(got.DSL.Profiles[0].Rules) != 1 {
-		t.Errorf("a rejected rule was persisted anyway: %d rules", len(got.DSL.Profiles[0].Rules))
-	}
-	// An unparseable price must be refused too, not silently priced at zero.
-	if rr := rulesDo(t, g, http.MethodPost, `{"profile":"p","rule":{"url":"https://y/v1","mode":"token",
-      "models":{"m":{"prompt":"free","completion":"0.6"}}}}`); rr.Code != http.StatusBadRequest {
-		t.Errorf("non-numeric price accepted with %d, want 400", rr.Code)
-	}
-}
-
-// TestBillingRulesPersistAndReachThePlugin is the point of the whole
-// endpoint, and mutation verification showed the earlier CRUD tests missed it
-// twice: dropping cfg.Save() and dropping the SetState injection both left
-// every test green. Both failures are invisible in a GET — the response echoes
-// the in-memory rules either way — so this test checks the two consequences
-// that actually matter:
-//
-//  1. persistence: the rule must be in the config file on disk, because an
-//     edit that evaporates on restart is a lie the UI told the operator;
-//  2. effect: the billing plugin must now PRICE the source (its published
-//     prices table has a non-zero entry), because a rule that is stored but
-//     never injected prices nothing at all.
-func TestBillingRulesPersistAndReachThePlugin(t *testing.T) {
-	g := rulesGateway(t)
-	cfg := g.core.Config()
-
-	seed := `{"billing":{"active":"p","profiles":[{"id":"p","rules":[
-      {"url":"https://free.example.com/v1","mode":"free"},
-      {"url":"https://api.commandcode.ai/v1","mode":"token",
-       "models":{"deepseek-v4.1-flash":{"prompt":"0.15","completion":"0.60"}}}]}]}}`
-	if rr := rulesDo(t, g, http.MethodPut, seed); rr.Code != http.StatusOK {
-		t.Fatalf("seed = %d: %s", rr.Code, rr.Body.String())
-	}
-
-	// 1. persisted to the file the process loaded from.
-	raw, err := os.ReadFile(cfg.Path)
-	if err != nil {
-		t.Fatalf("read config: %v", err)
-	}
-	if !strings.Contains(string(raw), "commandcode.ai") {
-		t.Errorf("the rule is not in config.yaml on disk — an edit that does not "+
-			"survive a restart:\n%s", raw)
-	}
-	reloaded, err := config.Load(cfg.Path)
-	if err != nil {
-		t.Fatalf("reload saved config: %v", err)
-	}
-	if reloaded.BillingDSL == nil || len(reloaded.BillingDSL.Profiles) == 0 {
-		t.Fatalf("saved config has no billing DSL: %s", raw)
-	}
-	// A round-trip through YAML must not lose the price strings.
-	found := false
-	for _, r := range reloaded.BillingDSL.Profiles[0].Rules {
-		if strings.Contains(r.URL, "commandcode") {
-			found = true
-			if m, ok := r.Models["deepseek-v4.1-flash"]; !ok || m.Prompt != "0.15" {
-				t.Errorf("price did not survive the YAML round-trip: %+v", m)
-			}
-		}
-	}
-	if !found {
-		t.Errorf("the token rule is missing after reload:\n%s", raw)
-	}
-
-	// 2. reached the plugin. Prices live in a separate Lua field from `state`,
-	// and the GET /state endpoint returns the published state together with
-	// the prices sidecar — which is exactly what the plugin's own UI reads.
-	state := g.core.Plugins().State("billing")
-	if state == nil {
-		t.Fatalf("billing plugin published no state")
-	}
-	rr := doReq(t, g, http.MethodGet, "/api/plugins/billing/state", "")
-	if rr.Code != http.StatusOK {
-		t.Fatalf("GET state = %d: %s", rr.Code, rr.Body.String())
-	}
-	var envelope struct {
-		Prices map[string]interface{} `json:"prices"`
-	}
-	if err := json.Unmarshal(rr.Body.Bytes(), &envelope); err != nil {
-		t.Fatalf("decode state envelope: %v", err)
-	}
-	prices := envelope.Prices
-	if prices == nil {
-		t.Fatalf("state payload carries no prices")
-	}
-	sources, ok := prices["sources"].(map[string]interface{})
-	if !ok {
-		t.Fatalf("prices has no sources map: %#v", prices)
-	}
-	cc, ok := sources["commandcode"]
-	if !ok {
-		t.Fatalf("the rule was saved but never injected: prices has %v, "+
-			"no commandcode entry. Every request on it stays unpriced.", keysOf(sources))
-	}
-	if cc == nil {
-		t.Error("commandcode price entry is nil")
-	}
-	// The free rule must price localzen, and the two must not be conflated.
-	if _, ok := sources["localzen"]; !ok {
-		t.Errorf("the free rule did not reach the plugin: %v", keysOf(sources))
-	}
-	// The injected price must be per-token, i.e. 0.15 USD/M => 1.5e-7.
-	// Asserting the exact magnitude catches a unit error (per-million vs
-	// per-token), which would be off by a factor of a million and still look
-	// like "a number".
-	models, ok := cc.(map[string]interface{})["models"].(map[string]interface{})
-	if ok {
-		if m, ok := models["deepseek-v4.1-flash"].(map[string]interface{}); ok {
-			p, _ := m["prompt"].(float64)
-			if p <= 0 || p > 1e-6 {
-				t.Errorf("injected prompt price = %v, want per-token (~1.5e-7); "+
-					"a per-million value here would over-bill by 1e6x", p)
-			}
-		}
-	}
-}
-
-func keysOf(m map[string]interface{}) []string {
-	out := make([]string, 0, len(m))
-	for k := range m {
-		out = append(out, k)
-	}
-	sort.Strings(out)
-	return out
-}
-
-// TestBillingRulesRequireAdmin keeps price rewriting behind the admin role,
-// like every other plugin write: prices are configuration, not user data.
-func TestBillingRulesRequireAdmin(t *testing.T) {
-	g := rulesGateway(t)
-	rec, err := g.core.CreateKey("viewer", "user", nil, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	userKey := rec.Key
-	for _, m := range []string{http.MethodPut, http.MethodPost, http.MethodDelete} {
-		req, _ := http.NewRequest(m, "/api/plugins/billing/rules", strings.NewReader(`{}`))
-		req.Header.Set("Authorization", "Bearer "+userKey)
-		req.Header.Set("Content-Type", "application/json")
-		if rr := newRecorderFor(t, g, req); rr.Code != http.StatusForbidden {
-			t.Errorf("user %s rules = %d, want 403", m, rr.Code)
-		}
-	}
-}

internal/gateway/chat.go

@@ -13,11 +13,11 @@ import (
 	"sync/atomic"
 	"time"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/lua"
-	"llmsproxy/internal/provider"
-	"llmsproxy/internal/scheduler"
-	"llmsproxy/internal/types"
+	"modelrouter/internal/config"
+	"modelrouter/internal/lua"
+	"modelrouter/internal/provider"
+	"modelrouter/internal/scheduler"
+	"modelrouter/internal/types"
 )
 
 // chatRequest mirrors the OpenAI chat completions request the gateway accepts.

internal/gateway/chat_cache_test.go

@@ -3,7 +3,7 @@ package gateway
 import (
 	"testing"
 
-	"llmsproxy/internal/types"
+	"modelrouter/internal/types"
 )
 
 // TestMergeUsageKeepsZeroCacheDetails locks the streaming counterpart of the

internal/gateway/chat_reasoning_test.go

@@ -3,7 +3,7 @@ package gateway
 import (
 	"testing"
 
-	"llmsproxy/internal/types"
+	"modelrouter/internal/types"
 )
 
 // TestRecordChatUsageKeepsReasoningTokens pins the reasoning-token plumbing.

internal/gateway/deploy_script_test.go

@@ -11,7 +11,7 @@ import (
 
 // deploy.sh is the only release-critical script with no automated coverage, and
 // its newest part — prune_adapter_backups — deletes directories under
-// /etc/llmsproxy. This exercises the function in a throwaway directory with the
+// /etc/modelrouter. This exercises the function in a throwaway directory with the
 // REAL backup-name shape.
 //
 // The shape matters and got this wrong twice: the guard is
@@ -45,7 +45,7 @@ func TestDeployPruneAdapterBackups(t *testing.T) {
 
 	// Redirect ONLY the base path. The logic under test must stay verbatim.
 	body = strings.Replace(body,
-		`local base="/etc/llmsproxy"`,
+		`local base="/etc/modelrouter"`,
 		`local base="`+base+`"`, 1)
 	// KEEP_BACKUPS must come from deploy.sh itself, NOT from a literal
 	// here: hardcoding 5 in the shim overrode whatever the script configured, so
@@ -207,7 +207,7 @@ func TestDeployBacksUpTheLiveFileBeforeOverwriting(t *testing.T) {
 //
 // 事实：backup 清理最初只处理 `adapters.bak.*`，而 `sync_plugins` 同样每次部署
 // 建一个 `plugins.bak.<时间戳>` —— 于是 adapters 的堆积被修好，plugins 继续涨。
-// 实测线上 plugins.bak 积累到 **36 个**、/etc/llmsproxy 涨到 192M。
+// 实测线上 plugins.bak 积累到 **36 个**、/etc/modelrouter 涨到 192M。
 //
 // 这是「两处只修一处」的老形状。判据钉住：prune_all_backups 必须同时裁两个前缀，
 // 且新增任何 .bak.<时间戳> 前缀时都要在这里补上（否则又是一个泄漏点）。
@@ -264,7 +264,7 @@ func TestDeployPruneCoversBothBackupPrefixes(t *testing.T) {
 	k := strings.Index(src, "prune_backups()")
 	kj := strings.Index(src[k:], "\n}\n")
 	fnBody := src[k : k+kj+3]
-	fnBody = strings.Replace(fnBody, `local base="/etc/llmsproxy"`, `local base="`+base+`"`, 1)
+	fnBody = strings.Replace(fnBody, `local base="/etc/modelrouter"`, `local base="`+base+`"`, 1)
 	keepRe := regexp.MustCompile(`(?m)^KEEP_BACKUPS=(\d+)`)
 	km := keepRe.FindStringSubmatch(src)
 	if km == nil {

internal/gateway/discover.go

@@ -11,7 +11,7 @@ import (
 	"strings"
 	"time"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // Model discovery: turn "here is a URL and a key" into a configured source.

internal/gateway/discover_api.go

@@ -5,7 +5,7 @@ import (
 	"encoding/json"
 	"errors"
 	"io"
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 	"net/http"
 	"strings"
 	"time"

internal/gateway/discover_api_test.go

@@ -9,7 +9,7 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // upstream that answers /v1/models with the real captured catalog

internal/gateway/discover_mutate_test.go

@@ -126,7 +126,12 @@ func TestRealMutants(t *testing.T) {
 		},
 		{
 			name: "nested metadata no longer read",
-			from: "\tcopyFields(m.Metadata)      // nested first",
+			// 锚点必须与 discover.go 逐字节一致：此前这里是 6 个空格，源码已
+			// 变成 2 个空格，于是这条变异永远"找不到锚点"——变异campaign 每次
+			// 都以 anchor not found 失败，而它本该验证"嵌套 metadata 被丢掉"能被
+			// 测试抓住。锚点漂移 = 这条变异从未真正跑过（和 issue #1 里
+			// `cc["models"]` 恒 false 的测试跳过是同一种失效：断言写了但没执行）。
+			from: "\tcopyFields(m.Metadata)  // nested first",
 			to:   "\tcopyFields(json.RawMessage(nil)) // mutant: nested metadata dropped",
 		},
 		{

internal/gateway/gateway_test.go

@@ -14,8 +14,8 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/core"
+	"modelrouter/internal/config"
+	"modelrouter/internal/core"
 )
 
 func mockUpstream() *httptest.Server {

internal/gateway/key_auto_api_test.go

@@ -9,8 +9,8 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/core"
+	"modelrouter/internal/config"
+	"modelrouter/internal/core"
 )
 
 // Per-key AUTO chain admin API. The behaviours pinned here are the ones that

internal/gateway/key_quota_api_test.go

@@ -9,8 +9,8 @@ import (
 	"strings"
 	"testing"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/core"
+	"modelrouter/internal/config"
+	"modelrouter/internal/core"
 )
 
 // adminGateway builds a gateway whose admin key can call /api/keys.

internal/gateway/key_quota_test.go

@@ -4,7 +4,7 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/config"
+	"modelrouter/internal/config"
 )
 
 // ---- data layer: per-key isolation ----

internal/gateway/key_quota_wiring_test.go

@@ -12,8 +12,8 @@ import (
 	"testing"
 	"time"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/core"
+	"modelrouter/internal/config"
+	"modelrouter/internal/core"
 )
 
 // quotaGateway builds a gateway with two user keys that differ only in their

internal/gateway/keys.go

@@ -6,8 +6,8 @@ import (
 	"net/http"
 	"strings"
 
-	"llmsproxy/internal/config"
-	"llmsproxy/internal/scheduler"
+	"modelrouter/internal/config"
+	"modelrouter/internal/scheduler"
 )
 
 // handleKeysAPI manages gateway keys: GET /api/keys (admin: all keys),

internal/gateway/messages.go

@@ -10,9 +10,9 @@ import (
 	"strings"
 	"time"
 
-	"llmsproxy/internal/provider"
-	"llmsproxy/internal/scheduler"
-	"llmsproxy/internal/types"
+	"modelrouter/internal/provider"
+	"modelrouter/internal/scheduler"
+	"modelrouter/internal/types"
 )
 
 // handleMessages serves the Anthropic Messages protocol (POST /v1/messages).