- Drop skill.NewManager from homed bootstrap; skills dir no longer core-managed
- Remove GetInjectedPrompt system-prompt injection (skills are not first-class)
- Delete internal/skill package, SkillAPI, webui /api/v1/skills, status skills block
- ConfigRegistry: plugin config tables now created only via RegisterDef; arbitrary
scope Set/Get no longer implicitly creates config_<name> tables (fixes stray
config_today_task table from SKILL directory name being used as a scope)
- handleChat now injects with source=webui so the LLM no longer sees
cli as the input channel
- subscribe to EventRawInput/EventAgentOutput to persist every
conversation turn from all channels (cli/qq/webui), replacing the
manual webui-only addChatMsg to avoid duplicates
- ChatMsg gains a source field; GUI shows a channel badge for
non-webui messages
- Six-phase plan complete: webui/cli/healthcheck/pluginmgr/clawhubadapter
now interact with the kernel exclusively via internal/sdk interfaces;
all Configure() calls and package-level global injection removed
- buildSDK in internal/plugin/registry.go is the single assembly point
- Add internal/sdk/events.go exporting event types/constants
- Fix ProviderManager cooldown sharing: LuaAdaptedProvider.Name() now
returns the source name instead of lua_<adapter>, so multiple sources
sharing an adapter (single script load via shared VM AdapterCache) no
longer share failure-cooldown state
- Verified: build/vet/tests green, deployed to homeagent.service with
full plugin capability testing via local OpenAI-compatible mock
All 5 adapters (deepseek, github, groq, mistral, openai) now:
- Guard against json.decode returning nil (null body)
- Use type() == "table" instead of truthy checks for table access
- Prevents 'attempt to index a non-table object(nil)' crashes
- Fix transform_response crash when json.decode returns nil (null body)
- Replace truthy checks with type() == "table" for safer field access
- Prevents 'attempt to index a non-table object(nil)' errors
- process.go: attach resp.ReasoningContent when building assistant messages
- deepseek.lua v2.1.0: remove last_reasoning closure hack; rely on
core-provided reasoning_content in messages
- openai.lua: strip reasoning_content from messages in transform_request
(not supported by OpenAI API)
- Manager now handles OC plugin loading via plugins/load JSON-RPC
instead of launching a separate simulator process
- Added 8 missing provider registration methods to manager
- Added explicit registeredChannels declaration in manager
- Provider dispatch now strips _provider suffix correctly (lookupType)
- Skip .simulator directory in skills scanning
- detect OC via package.json openclaw field (hasOCPackage)
- All registration notifications include plugin name for correct tool prefix
- Fix ToolRegistry/ProviderRegistry/ChannelRegistry to use plugin name
from notification data instead of hardcoded 'manager'
- test: add mockSettings, fix TestLoadOCPluginViaPluginStart panic
- registry.go: replace channelDevice+RegisterChannel with RegisterOutputChannel
handler routes output via sp.CallTool to JS manager outbound handlers
register {plugin}_read_{ch}_input tool for agent to poll buffered input
- plugin.go: handle channel_input notifications in translateAndRegister
buffer payload in channelInputBuf + InjectInterruptText to notify agent
- manager/main.js: registerChannel supports {plugin: ChannelPlugin} format
tools/call routes channel names to outbound.sendText/sendMedia
submitInput sends channel_input notification to Go side
- process.go: only emit resp.Content on the first tool call per batch,
subsequent assistant messages use empty content (serialized as null)
- provider.go: MarshalJSON outputs null content when empty with tool_calls
to comply with DeepSeek/OpenAI expected format
- output.go: update parameter interface (payload/type/meta) to match
tool definitions (uncommitted from previous refactor)
Previously the wrapper silently ignored json.Unmarshal errors and returned
nil, causing tools returning raw JSON strings (via C ABI bridge's
json.Marshal) to appear as empty map[] to the agent.
manager/main.js, pysimulator/main.py, and simulator/main.js are
production source files for the OpenClaw sidecar system, not build
artifacts. Remove them from .gitignore so fresh clones can make build
without manual stub creation.
- Redesign output_send__ tools: content JSON string -> structured
payload/meta/type params for LLM reliability
- executeOutputSendTool: route by type with capability check
- executeOutputSendHelp: show meta format + type enum
- Updated system prompt rules for new interface
- docToTriples: use jieba exact mode adjacent co-occurrence
- Unify vector space: Doc.Vector field, ContextToDoc vectorizer,
ReindexWithVectorizer on startup
- Add core.agent.embedding_model_path to seedDBValues and seedCoreDefs
- Wire cfgReg.GetString in main.go to AgentConfig.EmbeddingModelPath
- Follow existing core.agent.* ConfigRegistry pattern
