https://github.com/buvis/claude-checkup
Health-check and audit skills for Claude Code
https://github.com/buvis/claude-checkup
Last synced: about 2 months ago
JSON representation
Health-check and audit skills for Claude Code
- Host: GitHub
- URL: https://github.com/buvis/claude-checkup
- Owner: buvis
- License: mit
- Created: 2026-05-11T07:25:22.000Z (3 months ago)
- Default Branch: master
- Last Pushed: 2026-05-11T14:41:26.000Z (3 months ago)
- Last Synced: 2026-05-11T16:37:15.201Z (3 months ago)
- Language: Python
- Size: 55.7 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE
Awesome Lists containing this project
README
# Claude Checkup
[](https://github.com/buvis/claude-checkup/blob/master/LICENSE)
A health-check toolkit for [Claude Code](https://claude.ai/code). Thirteen focused audit skills plus a single orchestrator that runs them all, prints a dashboard, and produces a prioritized remediation plan.
## What it audits
| Skill | Catches |
|-------|---------|
| `audit-security` | Hardcoded secrets, loose permission patterns, hook injection, risky MCP servers |
| `audit-permissions` | Permission sprawl, unused grants, escalations |
| `audit-hooks` | Missing scripts, non-executable hooks, silent failures, slow hooks |
| `audit-settings` | Conflicts across global / project / local scopes, redundant overrides |
| `audit-mcp-health` | Disconnected MCP servers, stale config, last-used tracking |
| `audit-plugins` | Stale cached plugin versions, unused installs, disk reclaimable |
| `audit-memory` | Orphan memories, missing entries in `MEMORY.md` index |
| `audit-skills` | Skill structural validation, frontmatter, trigger patterns |
| `audit-rules` | Conflicts, shadowing, redundancies, staleness across rule files |
| `audit-context` | Per-component token overhead, cache classification |
| `audit-sessions` | Patterns, anomalies, unused skills across past sessions |
| `audit-project-orphans` | Stale project configs in `~/.claude/projects/` |
| `audit-claude-config` | **Orchestrator** — runs them all, prints dashboard, builds remediation plan |
## Install
Two commands inside Claude Code:
```
/plugin marketplace add buvis/claude-plugins
/plugin install claude-checkup@buvis-plugins
```
Restart Claude Code, then run `/audit-claude-config` to get a full health report.
### Update
```
/plugin update claude-checkup@buvis-plugins
```
### Alternative: install directly from this repo
```
/plugin marketplace add buvis/claude-checkup
/plugin install claude-checkup@claude-checkup
```
## Usage
```
audit my claude config # run everything, build remediation plan
audit security # security-only audits
audit health # health-only audits
audit efficiency # efficiency audits (slower)
```
Or invoke any individual audit by name, e.g. `audit hooks`, `audit memory`, `audit skills`.
The orchestrator saves a dated report to `dev/local/audit-results/{YYYY-MM-DD}.md` and diffs against the previous report (new findings, resolved findings, unchanged).
## Severity grading
Each finding gets a severity, prioritized in the remediation plan:
- **CRITICAL** — fix now (data loss, secret exposure, broken hooks)
- **HIGH** — fix this week (permission escalations, conflicting rules)
- **MEDIUM** — fix when convenient (cleanup opportunities)
- **LOW** — backlog (style, minor consistency)
## Requirements
- Claude Code with plugin support
- `python3` on PATH (helper scripts in `audit-security` and `audit-sessions`)
- Optional: [warden](https://github.com/buvis/claude-warden) — if installed, the orchestrator also calls `/warden:review-decisions`
## License
MIT