Skip to content
Back to search

skills-janitor

Tinder for your Claude Code skills, subagents and MCP servers. Swipe away what wastes context, scan for prompt injection, get honest token costs. Fixes what it finds and actually deletes. Zero dependencies, Claude Code + Codex.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/khendzel--skills-janitor/badge.svg)](/listings/khendzel--skills-janitor)

HTML

<a href="/listings/khendzel--skills-janitor"><img src="/api/listings/khendzel--skills-janitor/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

/plugin marketplace add khendzel/skills-janitor
/plugin install skills-janitor

Or via [skills.sh](https://skills.sh):

npx skills add khendzel/skills-janitor

Or clone directly:

git clone https://github.com/khendzel/skills-janitor ~/.claude/skills/skills-janitor

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 28 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "name": "skills-janitor",
  3. "owner": {
  4. "name": "Krzysztof Hendzel",
  5. "url": "https://github.com/khendzel"
  6. },
  7. "metadata": {
  8. "description": "Tinder for your Claude Code skills, subagents and MCP servers. Swipe away what wastes context, scan for prompt injection, get honest token costs. Fixes what it finds and actually deletes. Zero dependencies, Claude Code + Codex.",
  9. "version": "1.8.0",
  10. "repository": "https://github.com/khendzel/skills-janitor"
  11. },
  12. "plugins": [
  13. {
  14. "name": "skills-janitor",
  15. "description": "6 skill hygiene tools: health report, auto-fix + prune, value (honest token split + usage), security scan (prompt injection + malicious patterns), discover (search + precheck), swipe (interactive triage).",
  16. "source": "./",
  17. "strict": false,
  18. "skills": [
  19. "./skills/janitor-report",
  20. "./skills/janitor-fix",
  21. "./skills/janitor-security",
  22. "./skills/janitor-value",
  23. "./skills/janitor-discover",
  24. "./skills/janitor-swipe"
  25. ]
  26. }
  27. ]
  28. }
README.md · README · 162 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Skills Janitor
  2. > Tinder for your Claude Code skills. Swipe through your collection and delete what's wasting context, in seconds.
  3. Works with **Claude Code** and **OpenAI Codex**. 6 commands, zero dependencies.
  4. ![/janitor-swipe — swipe keep / delete / skip through every installed skill](janitor-swipe-demo.gif)
  5. > **Status (v1.8, September 2026): maintenance mode. Claude Code now does the core of this natively.**
  6. >
  7. > When this project started, nothing in Claude Code told you which skills were eating your context. Now it does:
  8. >
  9. > - [`/skill-doctor`](https://code.claude.com/docs/en/skills#find-unused-skills) shows what each skill costs in context and how often it gets used, and flags the ones never invoked
  10. > - [`/doctor`](https://code.claude.com/docs/en/commands) finds unused skills, MCP servers and plugins against their context cost, and fixes them after asking
  11. > - `/doctor prompt-audit` audits your CLAUDE.md files, skills, agents and commands for outdated or conflicting instructions
  12. > - `/skills` sorts by token count (`t`) and hides a skill with `Space`
  13. >
  14. > If you only use Claude Code and only want to trim context, use those. They see real usage from inside the harness, which a plugin never can.
  15. >
  16. > The janitor stays useful for what still isn't built in: **security scanning** ([`/janitor-security`](#security-scan-v16), plus a pre-install check in `/janitor-discover`), **duplicate detection** across skills, **OpenAI Codex** support, and **actually deleting** broken symlinks and dead skill folders. Bug fixes and PRs are still welcome. No new features are planned.
  17. >
  18. > Thanks to everyone who installed it, filed issues and sent fixes: #4 of the day on Product Hunt, 120+ stars, and PRs from [@nshonda](https://github.com/nshonda), [@lulzpid](https://github.com/lulzpid) and [@jerryhuangzq-lang](https://github.com/jerryhuangzq-lang). We were early, and now it's native. That's a good ending for a janitor.
  19. Scans every place a skill lives: user, project, codex, and every skill installed via `/plugin install` — plus your subagents and MCP servers. Surfaces duplicates, broken symlinks, unused skills, and connected-but-never-called MCP servers cluttering your context. Usage counts come from real session transcripts, including skills Claude auto-triggered.
  20. ## Commands
  21. | Command | What it does |
  22. |---|---|
  23. | `/janitor-report` | Health check: inventory, duplicates, broken skills. `--brief` for inventory only. |
  24. | `/janitor-fix` | Auto-fix issues. `--prune` removes broken symlinks and empty dirs. |
  25. | `/janitor-value` | Honest token costs (always-loaded descriptions vs on-demand bodies) + usage, skills and subagents. |
  26. | `/janitor-security` | Heuristic scan for prompt injection, hidden instructions, and dangerous script patterns. (v1.6+) |
  27. | `/janitor-discover` | Search GitHub for skills, or check a URL before installing — now including a pre-install security scan. |
  28. | `/janitor-swipe` | Interactive TUI — swipe keep/delete/skip through skills AND MCP servers, sorted most-likely-waste first. (v1.4+) |
  29. Each has its own slash command. Or use natural language: *"check my skills"*, *"which skills are wasting context?"*, *"find an n8n skill"*.
  30. ## Install
  31. ```
  32. /plugin marketplace add khendzel/skills-janitor
  33. /plugin install skills-janitor
  34. ```
  35. Or via [skills.sh](https://skills.sh):
  36. ```bash
  37. npx skills add khendzel/skills-janitor
  38. ```
  39. Or clone directly:
  40. ```bash
  41. git clone https://github.com/khendzel/skills-janitor ~/.claude/skills/skills-janitor
  42. ```
  43. ## Security scan (v1.6)
  44. A skill is text your agent trusts. Public research found prompt injection in roughly a third of tested community skills — so the janitor now scans for the known bad shapes: instruction-override phrases, "don't tell the user" directives, instructions hidden in HTML comments or zero-width unicode, smuggled base64 payloads, and scripts that pipe the network into a shell or read credential stores.
  45. ```
  46. /janitor-security # audit everything installed
  47. /janitor-discover <url> # overlap + security check BEFORE installing
  48. ```
  49. Verdicts are honest heuristics (PASS / REVIEW / RISK): a RISK means "read this before trusting it", not "malware". Calibrated for low noise — on a real 178-skill machine it flags 2, both genuinely worth reading.
  50. ## Swipe through your skills (v1.4)
  51. Tinder-style triage for your skill collection. The deck is sorted heaviest-and-least-used first, so most users hit `← delete` through the top 5–10 cards and quit before reviewing everything.
  52. ```
  53. !bash ~/.claude/skills/skills-janitor/scripts/swipe.sh
  54. ```
  55. (The `!` prefix runs in your terminal, not the Claude Code Bash tool — the TUI needs a real interactive stdin.)
  56. Controls: `←` delete, `→` keep, `↓` skip, `u` undo, `i` inspect full description, `q` quit.
  57. Plugin skills are flagged for review (you can't `rm` individual plugin skills — they belong to a plugin). User-scope skills stage for actual deletion, applied on `y` confirmation at the end.
  58. Since v1.7 the deck also includes your MCP servers — see [MCP server triage](#mcp-server-triage-v17).
  59. ## MCP server triage (v1.7)
  60. Skills aren't the only thing renting space in your context. Every connected MCP server loads
  61. its tool schemas on every request, whether you call it or not.
  62. The janitor inventories every configured server — user `~/.claude.json`, per-project entries,
  63. project `.mcp.json`, and plugin-bundled — then cross-references real usage from your session
  64. transcripts (`mcp__server__tool` records):
  65. ```
  66. === Skills Janitor - MCP Servers ===
  67. Usage window: last 8 weeks of session transcripts
  68. Server Scope Calls Tools Last Used Origin
  69. ──────────────────────────── ──────────── ────── ────── ─────────── ────────────
  70. design-tool mcp-user 0 0 never ~/.claude.json
  71. deploy-tool mcp-plugin 0 0 never plugin:deploy
  72. cms mcp-project 41 4 2026-07-02 ~/.claude.json
  73. --- Unused in 8 weeks (2) ---
  74. design-tool (mcp-user) — configured in ~/.claude.json
  75. deploy-tool (mcp-plugin) — configured in plugin:deploy
  76. ```
  77. No invented token numbers — MCP schemas live server-side, so the janitor reports only what it
  78. can prove: where the server is configured, how many distinct tools you actually called, how
  79. often, and when last. Servers seen in transcripts but no longer configured are listed
  80. separately.
  81. Unused servers rank high in the swipe deck. Swiping one left removes the entry from its config
  82. file with a timestamped `.bak` backup. Plugin-bundled servers are flagged for plugin review
  83. instead.
  84. ## Duplicate detection
  85. The duplicate detector flags cross-scope overlaps, including plugins that re-implement a skill
  86. you already had standalone:
  87. ```
  88. === Skills Janitor - Duplicate Detection ===
  89. --- Description Overlap (Jaccard > 30%) ---
  90. [98%] marketing-seo-audit <-> marketing-skills:seo-audit
  91. Scopes: user / plugin
  92. [100%] marketing-content-strategy <-> marketing-skills:content-strategy
  93. Scopes: user / plugin
  94. ```
  95. Overlap is scored on descriptions (Jaccard), so it catches re-implementations that share no
  96. filename and live in different scopes.
  97. ## Upgrading from v1.2
  98. The five v1.2 aliases were removed in v1.5. Renames:
  99. | v1.2 | now |
  100. |---|---|
  101. | `/janitor-audit` | `/janitor-report --brief` |
  102. | `/janitor-usage` | `/janitor-value` |
  103. | `/janitor-tokens` | `/janitor-value` |
  104. | `/janitor-search` | `/janitor-discover` |
  105. | `/janitor-precheck` | `/janitor-discover <url>` |
  106. Full release notes: [CHANGELOG.md](CHANGELOG.md).
  107. ## Requirements
  108. Bash, Python 3, `curl`. No pip installs, no node modules.
  109. ## Contributing
  110. PRs welcome. Each command is self-contained in `skills/janitor-*/SKILL.md` plus a sibling script in `scripts/`.
  111. ## License
  112. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.