Run a 6-hat brutal LLM Council self-critique on any deliverable, session output, or status update. Surfaces what's actually missing, where the work is shallow, and what was tabled-for-later. Use BEFORE re-attacking a problem the maintainer has flagged twice.
安装
npx skills add naman10parikh/agentgrid --skill llm-council-critiquenpx skills add naman10parikh/agentgrid --skill llm-council-critique -a claude-code -gnpx skills add naman10parikh/agentgrid --skill llm-council-critique -a codex -ggemini skills install https://github.com/naman10parikh/agentgrid.git --path .claude/skills/llm-council-critiquegh skill install naman10parikh/agentgrid llm-council-critiquegit clone --depth 1 https://github.com/naman10parikh/agentgrid /tmp/naman10parikh__agentgrid
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/naman10parikh__agentgrid/.claude/skills/llm-council-critique ~/.claude/skills/llm-council-critique # Claude Code
cp -R /tmp/naman10parikh__agentgrid/.claude/skills/llm-council-critique ~/.agents/skills/llm-council-critique # Codex, Cursor, Gemini CLI, Copilot, OpenCode