Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsSkill Trigger Tuner: Audit and Rewrite SKILL.md Descriptions
Audits SKILL.md descriptions, finds trigger collisions between skills and rewrites each description as a "Use when / Do not use when" classifier you can measure.
$5
Skill Trigger Tuner: Audit and Rewrite SKILL.md Descriptions
Example session with this skill installed
Audit the three skills in this folder and tell me which ones overlap.
- Read your context and instructions
- Compiled the trigger tuner
Summary
Audited 3 skills. changelog-helper and release-notes collide (Jaccard 0.636, containment 0.875), and neither has a "Use when" or "Do not use when" clause. Output below is the real script output from the bundled demo.
Trigger audit
| skill | length | Use when | Do not use when | phrases | errors | warnings |
|---|---|---|---|---|---|---|
| changelog-helper | 98 | no | no | 0 | 0 | 3 |
| commit-message-helper | 258 | yes | yes | 4 | 0 | 1 |
| release-notes | 86 | no | no | 0 | 0 | 4 |
Collisions
| level | skill A | skill B | Jaccard | containment | shared words |
|---|---|---|---|---|---|
| collision | changelog-helper | release-notes | 0.636 | 0.875 | change, changelog, note, project, release, summary, write |
Thresholds are heuristics for spotting overlap; the agent's real choice depends on the model and the full prompt.
Next
I rewrite release-notes with the Scope–Signal–Fence method, then score the old and new versions on the demo test set and show the comparison table.
Connects securely to your tools. The creator never sees your data.
About this skill
Use when a skill never loads, the wrong skill loads, or a folder of skills has grown and you suspect overlap.
A standard-library Python audit checks every description against Anthropic's documented limits (1024 characters, reserved words in name), counts trigger phrases and flags colliding pairs with Jaccard and containment scores. The skill then rewrites each weak description with the Scope–Signal–Fence method and scores the old and new versions side by side on a should-fire / should-not-fire test set, using a transparent word-overlap proxy.
Do not use when you need a brand-new skill written from scratch, a review of the instructions inside the skill body, or marketplace listing copy.
What the scores mean
The evaluation is a word-overlap proxy for before/after comparison. It shows whether a description's wording covers the prompts you care about. It does not predict how often a particular model will load the skill.
Example prompts
- "My deploy skill never triggers. Here's the SKILL.md, fix the description."
- "Audit my .claude/skills folder and tell me which skills overlap."
- "Rewrite this description so it fires on 'write release notes' but not on commit messages, and show me before and after."
Independent
Not affiliated with or endorsed by Anthropic. Product names are used only to describe compatibility.
Scores are a word-overlap proxy for before/after comparison; they do not predict how often a particular model will load the skill. Symlinks are not followed during audits.
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
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 9 days ago
Requires