- Home
- Skills
- Prompt & Skill Engineering
- tidyagents
Works with the AI Agents you already use
tidyagents
Lint, score and safely tidy AI agent instruction files (AGENTS.md, CLAUDE.md, .cursorrules): finds stale paths, dead scripts, contradictions and bloat.
Free
tidyagents
Example session with this skill installed
Run tidyagents on this repo, tell me the score, and clean up what's safe to fix.
- Read your context and instructions
- Compiled the tidyagents
Summary
tidyagents scored CLAUDE.md
61/100, found 6 issues, and fix safely removed 3 of them, taking it to
76/100 (74 → 53 tokens). 3 issues need a human decision.
$ npx -y tidyagents
CLAUDE.md 61/100 (74 tokens)
✗ line 5 references `scripts/deploy.sh`, which does not exist [missing-path]
✗ line 6 runs script "typecheck", which is not in package.json [missing-script] fixable
⚠ line 7 "follow best practices" is too vague for an agent to act on [vague]
✗ line 8 contradicts line 3; keep one of them [contradiction]
⚠ line 9 repeats line 4 in the same section [duplicate] fixable
· line 11 3 blank lines in a row [blank-lines] fixable
Safe fixes applied (tidyagents fix)
| Line | Removed | Rule |
|---|---|---|
| 6 | Run npm run typecheck before pushing. | missing-script |
| 9 | duplicate Run npm run test before committing. | duplicate |
| 11-12 | extra blank lines | blank-lines |
| Metric | Before | After |
|---|---|---|
| Score | 61/100 | 76/100 |
| Tokens | 74 | 53 |
Needs your decision
- Line 5:
scripts/deploy.shis gone. Update the path, or delete the line if deploys moved. - Line 7: "follow best practices" is too vague. Name the concrete command or rule.
- Line 8: "Never use tabs" contradicts "Always use tabs" on line 3. Keep one.
Next steps
- Fix the three items above, then re-run
npx -y tidyagentsto reach 100. - Add the GitHub Action (
vishwakulkarni/tidyagents@v1,min-score: 80) so new rot fails the pull request.
Connects securely to your tools. The creator never sees your data.
About this skill
tidyagents lints, scores and tidies the instruction files your AI coding agents read every session: AGENTS.md, CLAUDE.md, GEMINI.md, SKILL.md, .cursorrules and copilot-instructions.md. Agent files rot quietly. Code moves, scripts get renamed, and the instructions keep pointing at things that no longer exist. This skill has your agent find that rot and clean up what is safe to clean. It checks for missing paths, dead npm scripts, contradictions, duplicates, vague rules and oversized files, then scores each file out of 100. The fix command only deletes lines it is sure about, shows a diff first, and flags anything uncertain instead of removing it. Zero dependencies, no network calls, no model calls, deterministic scores that work as a CI gate. In a scan of 67 popular open-source repos, 27% sent their agent to files that no longer exist. Source: github.com/vishwakulkarni/tidyagents (Apache 2.0).
Skips gitignored paths and example text to avoid false positives. Does not verify external network resources. Node >= 18.3 required.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Requires