kimandClaude Sonnet 4.6
81ae83a8fc
.gitignore에서 .smallclaw/skills/ 제외 해제. lawyer, psychiatrist, multi-agent-orchestrator SKILL.md 추가. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
1.2 KiB
1.2 KiB
name, description, emoji, version
| name | description | emoji | version |
|---|---|---|---|
| Multi-Agent Orchestrator | Enables a secondary AI model to advise the primary when it gets stuck, fails repeatedly, or needs upfront planning. | AI | 1.0.0 |
Multi-Agent Orchestration Active
You have access to a secondary AI advisor through the request_secondary_assist tool.
When to call request_secondary_assist
Call it proactively when:
- You need an upfront plan before starting a complex multi-file task -> use
mode: "planner" - You have failed the same action 2+ times -> use
mode: "rescue" - You are unsure which files to edit or what search queries to use -> use
mode: "planner" - You detect you are going in circles -> use
mode: "rescue"
What the advisor returns
The advisor returns a structured action plan with:
- next_actions: exactly what to do next, in order
- hints: specific search queries, file paths, tool arguments
- stop_doing: patterns to avoid
- risk_note: warnings about dangerous edits
Rules
- Follow the advisor's
next_actionsin order. - Use the exact
hintsprovided (search queries, file paths, etc). - Do not call
request_secondary_assistagain within 3 steps of the last call. - The advisor advises only - you still execute all tool calls yourself.