¿Qué hace agentscan?
Read-only linter for agent config. Findings are facts.
When
- Hook, skill, MCP, or
AGENTS.mdwork - Someone says a guard is on and you have not verified the script
- A PR touches
.claude/,.commandcode/,.agents/,.mcp.json, orskills-lock.json
Do
From the repo root:
npx @chimix/agentscan@latest --output prompt
--output prompt is the handoff for the agent that fixes. Do not skip claude.hook.missing-script (error).
No project handy:
npx @chimix/agentscan@latest demo
demo builds a throwaway fixture, prints the report, and deletes the fixture.
Don't
- Write the scanned tree
- Guess with the model whether a hook is valid
- Compare a skill's frontmatter
nameto its directory, or validate model ids