harness-doctor
Diagnose the harness itself. Reports version drift vs upstream, plugin staleness, MCP version pins, hook log size, memory growth, settings.json size. Run monthly or when behavior feels off.
Install / Use
npx skills add qyndex/claude-initInstalls into whichever agent you are using.
Claude Commands
Claude Code slash commands
Quality Score
Category
Development & EngineeringSupported Platforms
Our assessment of harness-doctor
harness-doctor scores 62/100 on our quality scale, 4042nd of 4,611 Development & Engineering skills we index.
Its Claude Commands is 2.6 KB long, well organised into 11 sections with 1 code example: a solid amount of guidance for an agent.
It has no GitHub stars yet, so there is no community track record; judge it on its content.
Maintenance, license and trust
- We could not determine when the repository was last updated.
- Our last check on 2026-10-20 found the source still online.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 68/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-24. Automated pattern scan on 2026-09-24. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
harness-doctor compared with similar skills
All 4 of these similar skills score higher than harness-doctor; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| harness-doctor (this skill)by qyndex | 62 | 0 | — | Claude Commands |
| claude-memby thedotmack | 100 | 113.6k | today | CLAUDE.md |
| taste-skillby Leonxlnx | 100 | 110.8k | 1d ago | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 109.3k | today | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 91.0k | 11d ago | CLAUDE.md |
Frequently asked questions
- How do I install harness-doctor?
- Run
npx skills add qyndex/claude-init. The install tabs above show the steps for each supported agent. - Which AI agents does harness-doctor work with?
- It is written for Claude Code, as a Claude Commands file. Other agents that read the same format can often use it too.
- Is harness-doctor safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 68/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is harness-doctor still maintained?
- We could not determine when the repository was last updated.
Skill content
View source on GitHubdescription: Diagnose the harness itself. Reports version drift vs upstream, plugin staleness, MCP version pins, hook log size, memory growth, settings.json size. Run monthly or when behavior feels off. argument-hint: "[--fix]" allowed-tools: Read, Glob, Grep, Bash, WebFetch disable-model-invocation: true
/harness-doctor — Self-diagnostic
ROOT="$(cd .claude/.. && pwd)"
echo "# Harness Doctor — $(date -Iseconds)"
echo
# 1. Version check
echo "## Version"
if [ -f .claude/VERSION ]; then
cat .claude/VERSION
else
echo "⚠ .claude/VERSION missing — cannot determine harness version"
fi
# 2. Compare to upstream (if origin set)
if [ -f .claude/VERSION ]; then
upstream=$(grep 'source=' .claude/VERSION | cut -d= -f2)
if [ -n "$upstream" ]; then
echo
echo "## Upstream drift"
echo "Source: $upstream"
latest=$(claude -p "WebFetch $upstream/raw/main/.claude/VERSION and return the harness_version field" --bare 2>/dev/null)
echo "Latest: $latest"
fi
fi
# 3. Plugin staleness
echo
echo "## Plugin staleness"
if command -v claude >/dev/null; then
claude -p "/plugin list with last-update dates" --bare 2>/dev/null | head -30
fi
# 4. MCP version pins
echo
echo "## MCP server pins"
echo "Pinned to @latest (supply-chain risk):"
grep -h '@latest' .mcp.json 2>/dev/null | head -10
echo
echo "Pinned to specific versions (good):"
grep -h '@[0-9]' .mcp.json 2>/dev/null | head -10
# 5. Hook log size
echo
echo "## Hook log size"
if [ -d .claude/hooks/.log ]; then
du -sh .claude/hooks/.log/* 2>/dev/null | sort -rh | head -10
fi
# 6. Memory file size
echo
echo "## Memory file sizes"
for f in .claude/memory/MEMORY.md .claude/memory/decisions/*.md .claude/memory/incidents/*.md; do
[ -f "$f" ] && wc -l "$f" 2>/dev/null
done | sort -rn | head -20
# 7. Settings.json size
echo
echo "## Settings.json metrics"
echo "Lines: $(wc -l < .claude/settings.json)"
echo "Allow patterns: $(jq -r '.permissions.allow | length' .claude/settings.json)"
echo "Ask patterns: $(jq -r '.permissions.ask | length' .claude/settings.json)"
echo "Deny patterns: $(jq -r '.permissions.deny | length' .claude/settings.json)"
# 8. Validate
echo
echo "## Validate"
bash .claude/scripts/validate.sh 2>&1 | tail -3
# 9. Recommend
echo
echo "## Recommendations"
echo "- If MCP @latest pins → run /audit-plugins to pin"
echo "- If MEMORY.md > 200 lines → run /dream"
echo "- If settings.json > 500 lines → /audit-settings to prune unused entries"
echo "- If upstream drift → git remote update && consider pulling harness updates"
Recurring scan
Add to dream-cron so this runs daily and surfaces issues at session-start via additionalContext.
$ARGUMENTS
Related Skills
claude-mem
113.6kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
taste-skill
110.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Agent-Reach
109.3kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
Understand-Anything
91.0kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
