| name | doctor |
| description | Validate a Claude Code install and the current project setup. Use when Claude needs to audit ~/.claude wiring, helper scripts, hooks, agents, dependencies, or optional MCP/statusline setup before troubleshooting. |
| argument-hint | [quick|full] |
| allowed-tools | ["Bash","Read","Grep","Glob"] |
| disable-model-invocation | true |
| metadata | {"version":"2.0","effort":"medium","auto-invocable":false,"category":"workflow","compatible-claude-code":{"when_to_use":"When troubleshooting a Claude Code/Kilo setup","allowed-tools":["Bash","Read","Grep","Glob"]}} |
Doctor Skill
Usage: /doctor [quick|full]
Run the Claude doctor script and summarize the results.
Process
- Parse
$ARGUMENTS. Default to quick.
- Prefer the installed script:
py ~/.claude/scripts/claude_doctor.py --mode "<mode>" --project-root "$PWD"
- If the installed script is missing but this repo contains the source, fall back to:
py claude-config/scripts/claude_doctor.py --mode "<mode>" --project-root "$PWD"
- Summarize the output in priority order:
- blocking install failures first
- missing hook or helper-script wiring second
- optional MCP/statusline/setup gaps last
Output Rules
- Quote the exact missing file, directory, or settings key when the script reports one.
- Keep the summary short; users can rerun the script for the full raw audit.
- Do not edit anything automatically unless the user explicitly asks for fixes.