Claude Ecosystem Hygiene

Claude Code Ecosystem Hygiene

Ten complementary skills across three tracks. The **Claude-ecosystem track** (5 plugins) audits, measures, cleans, keeps consistent, and curates the context budget of your `~/.claude/` stack: identify what's HOT vs DORMANT, measure whether the HOT artifacts actually improve task outcomes, prune what doesn't pull its weight, keep project docs in sync with the lessons that supersede them, and trim the runaway skills-catalog token tax (`context-police`) without deleting a skill. The **project-quality track** (`test-effectiveness-auditor` + `repo-hygiene`) applies the same "measure don't guess, clean before you ship" discipline to your project itself — measure how many real bugs your test suite actually catches, and clean the repo (stray data files, hardcoded paths, client data) before each PR. The **authoring track** (`skill-portfolio-repo-placement-scan` + `claude-plugin-repo-ci-release`) maintains the skills and plugin repos you publish — decide which repo each authored skill belongs in (ADD / UPDATE-with-direction / CROSS-LINK), and wire CI validation + release-on-version-bump into plugin/marketplace repos so GitHub Releases never drift behind the shipped version.

[](LICENSE) [](https://github.com/wan-huiyan/claude-ecosystem-hygiene/commits) [](https://claude.com/claude-code)


The audit → measure → clean → stay-consistent pipeline

┌─────────────────┐   ┌──────────────────┐   ┌────────────────┐   ┌───────────────────┐
│ ecosystem-audit │──▶│ claude-code-ab-  │──▶│ memory-hygiene │──▶│ doc-freshness-    │
│                 │   │ harness          │   │                │   │ reverse-lint      │
│ which artifacts │   │ do the HOT on