Install
openclaw skills install @wangyuchen5110/skills-doctorAudit the skills OpenClaw loads and get a measured plan to cut their per-step token cost.
openclaw skills install @wangyuchen5110/skills-doctorEvery eligible skill costs the prompt ~97 characters plus its name, description and location - on every step. Audit that cost before adding more skills.
SKILL.md references files that may have moved.npx -y openclaw-skills-doctor doctor # paths, roots, MCP + skill install state
npx -y openclaw-skills-doctor check # findings + per-step token cost
npx -y openclaw-skills-doctor plan # what to trim / hide, in tokens
npx -y openclaw-skills-doctor scan # every skill: source, precedence, cost
check is read-only. Add --json for structured output, --html report.html
for a shareable report, --budget <n> to change the 1200-token/step target.
plan; prefer trimming descriptions (keeps discoverability) over hiding.fix --dry-run then fix archives the losers
into .skills-doctor-archive/ (reversible).references/, scripts/ or assets/ path does not
resolve next to the SKILL.md. Fix the path; the file may live one level up.references/.npx -y openclaw-skills-doctor fix --dry-run # what would be archived/hidden
npx -y openclaw-skills-doctor fix # archive shadowed duplicates (reversible)
npx -y openclaw-skills-doctor fix --hide old-skill # autoinvoke: false, still loadable via @id
Never edit the top-level openclaw.json by hand from here - install/uninstall
own the mcp.servers.skills-doctor entry.
When the MCP server is registered (npx -y openclaw-skills-doctor install), call
the tool skills-doctor__skills_doctor instead of shelling out. It returns the
same markdown report and accepts directory, budget, lang and html.