Claude Code Doctor — Security skill for Claude Code
Health checkup for your Claude Code setup: read-only parallel audit, A-E grades with radar chart, red flags, HTML dashboard.
How to install Claude Code Doctor
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open kgraph57/claude-code-doctor and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Code Doctor does
Health checkup for your Claude Code setup: read-only parallel audit, A-E grades with radar chart, red flags, HTML dashboard. Battle-tested with Claude Fable 5 / Opus. Zero changes before you approve.
Alternatives in Security
- T3mp3st — autonomous red teaming platform; multi-agent offensive-security meta-harness 5.7k ★
- Data Viz Audit — Audit a data visualization for chart type selection, accessible color palette, annotations, and anti-patterns 309 ★
- Rolecraft — The security-first skill manager for AI agents — every install runs a security scan 75 ★
README
🇯🇵 **日本語版READMEがあります** → [README.ja.md](README.ja.md)
**English** | [日本語](README.ja.md)
[](LICENSE) [](https://github.com/kgraph57/claude-code-doctor/pulls) [](https://github.com/kgraph57/claude-code-doctor/actions/workflows/test.yml) [](https://claude.com/claude-code)
I let Claude Code audit its own setup.
It came back with 104 findings.
*Find the context tax, dead permissions, MCP bloat, and zombie automations hiding in your Claude Code setup.*
Current release: **v0.13.0** — Example Manifest validation, Renderer Bug validation, Contributed Report validation, Linux beta probe plan, 60-second walkthrough generator, Windows beta probe plan, Cross-Harness Checkups, Community Domain Packs, CI Budget Gate, and Diff Mode. See [CHANGELOG.md](CHANGELOG.md).
The 104-Finding Checkup
Claude Code Doctor is a read-only checkup for the AI-workspace layer: the context, permissions, tools, skills, automations, and evidence trails that shape what your coding agent sees before it writes a single line. Patient zero was my own two-year Claude Code setup:
| What | Found |
|---|---|
| Always-on context loaded every session | 45,000 tokens (22.5% of the window, gone before work starts) |
| Permission allow-list entries | 804, of which 77 pointed at paths that no longer exist |
| MCP tools injected every session | 63, including two servers that fully overlap |
| Zombie automations failing silently every morning | 12 |
| Skills with colliding trigger words | several clust |
Related Skills
Fable5 Audit Lite
Scan your Claude Code skills for patterns that silently break on Claude Fable 5 (silent Opus fallbacks, API re
Audit Chart
Audit the Helm chart for sync with operator code — values keys, CLI flags, CRD versions, template correctness,
MCP Approve
Run the Zero-Trust Validation Protocol to audit and approve an MCP package for installation
Claude Mission Control
Real-time command center for Claude Code agents. Monitor all agents from one dashboard, send instructions, tra
Claude Doctor Skill
46 automated checks across 6 layers. Project health audit skill for Claude Code. Security first. Zero dependen
Memory Checkup
Audit and repair an AI agent's memory corpus for consistency: stale numbers, dangling references, dead daily f
Related Agents
Campaign Operator
Deploys campaigns to Meta Ads (via the meta ads CLI), Google Ads (a real hybrid REST :mutate write flow + goog
Battle Test
Bundle 24 — full-corpus zero-FP evidence rule (row 15 + Challenger obligation)
Lead Judge
Use to work the Lead Radar queue: read each lead, agree or disagree with its score, approve the real ones with
