菜单
Switch or show the AI model for the current session. Usage: /model [model-name|default|list]
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Check Claude Brain status — shows whether hooks, MCP server, and memory are working
Analyzes code for bugs, security vulnerabilities, performance problems, and style issues. Use when reviewing a PR, reading a diff, auditing a file, or asked to check, critique, inspect, or audit code. Outputs a structured markdown report with severity ratings.
Breaks a complex multi-step request into subtasks and routes each to the appropriate skill. Use when given a high-level goal that requires multiple steps, a multi-phase workflow, or when asked to plan and execute a project, pipeline, or sequence of tasks. Outputs a step plan with skill assignments and handoff contracts.
Condenses any text into a structured summary with key points and a one-sentence takeaway. Use when asked to summarize, TLDR, condense, distill, or recap text, articles, transcripts, meeting notes, or documents. Outputs an overview paragraph, bullet-point key points, and a single takeaway line.
Show token usage and estimated cost for this session
Export the current conversation history to a formatted transcript. Usage: /export [format]
| name | model |
| description | Switch or show the AI model for the current session. Usage: /model [model-name|default|list] |
The user wants to view or change the AI model used for this session.
/model → show the current model override (or "provider default" if none set)/model list → same as above, plus list available models/model claude-opus-4-6 → switch to that model for this session immediately/model default or /model reset → revert to the provider's configured default modelPer-session model switching is fully supported. When you run /model <name>:
<name> instead of the provider default./model default similarly clears the session and reverts.Anthropic
claude-opus-4-6 — most capable, best for complex tasksclaude-sonnet-4-6 — fast and smart, best balanceclaude-haiku-4-5 — fastest, cheapestOpenAI
gpt-4o — flagship multimodal modelgpt-4o-mini — fast and cheapo3 — extended reasoningo4-mini — fast reasoninggemini-2.5-pro — Google's most capablegemini-2.5-flash — fast and cheapGroq
llama-3.3-70b-versatile — fast inferencemixtral-8x7b-32768 — large contextOllama (local)
llama3.3, mistral, qwen2.5, phi4, etc. — whatever you have pulled locallyThe /model command is handled directly by BrainClaw before this skill runs.
If the user asks which models are available, list the options above.
If the user asks whether per-session switching is supported: yes, it is — no restart needed.
/model default reverts it.