- Home
- Skills
- Agents & Orchestration
- Skill Health Scanner
Skill Health Scanner
Instantly diagnose any skill or prompt and get a clear, prioritized report on what’s wrong and how to fix it — across any agent.
Free
Works with the AI Agents you already use
Skill Health Scanner
Example session with this skill installed
I have this skill for writing good PR descriptions. Can you scan it?
name: "Good PR Descriptions"
description: "Write clear pull request descriptions"
version: 1.0.0
Good PR Descriptions
Write a good pull request description for the changes in the current branch.
Agent Operating Procedure
- Look at the git diff
- Understand what changed
- Write a description
- Read your context and instructions
- Compiled the health scanner
Skill Health Report
Skill Name
Good PR Descriptions
Target Models
Universal / Cross-Agent (no specific models provided)
Overall Score & Verdict
4/10 — 🔴 Critical Issues
Dimension Scores
- Clarity of Purpose: 6/10 — The goal is understandable but very generic.
- Trigger Conditions: 2/10 — No positive or negative examples. Extremely vague.
- Instruction Quality: 3/10 — Extremely high-level. No structure or guardrails.
- Output Contract: 2/10 — No template or required format.
- Safety & Guardrails: 8/10 — No obvious risks.
- Cross-Agent Portability: 3/10 — Heavily relies on implicit understanding of “git diff”.
Model Compatibility Notes
- Claude 3.x: ⚠️ Needs Work — Very likely to produce vague or overly long descriptions.
- GPT-4 family: ❌ Critical Issues — Will probably hallucinate a PR description instead of following a process.
- Gemini 1.5+: ❌ Critical Issues — No structure for the model to follow.
- Local / Smaller Models: ❌ Critical Issues — Far too underspecified.
Top Issues (Prioritized)
-
No output template
- Why it matters: Without a required structure, every run produces different quality and format.
- How to fix: Add a mandatory markdown template with specific sections.
-
Completely missing trigger conditions
- Why it matters: The skill has no idea when it should or should not be used.
- How to fix: Add clear “When to Use” and “When NOT to Use” sections with examples.
-
No Agent Operating Procedure
- Why it matters: The current instructions are too vague for reliable execution.
- How to fix: Break the process into explicit, numbered phases with verification steps.
Recommended Fixes
Quick Wins (Under 10 minutes)
- Add a required PR description template.
- Write “When to Use” and “When NOT to Use” sections with 2–3 examples each.
Structural Improvements (10+ minutes)
- Turn the current 3-step list into a proper 5-phase Agent Operating Procedure.
- Add a “Verification” step after drafting the description.
Next Steps (Optional)
You can implement these fixes manually, or ask for recommendations on deeper tools if you want systematic help.
Connects securely to your tools. The creator never sees your data.
About this skill
Instantly diagnose why your skills and prompts underperform
You’ve spent time writing a skill or prompt, but the results are inconsistent,
it only works in one agent, or it fails to trigger reliably.
Skill Health Scanner gives you a fast, structured diagnosis
of any SKILL.md or system prompt — no matter which agent it was written for.
What you get
- Model-aware analysis — See how your skill performs across Claude, GPT-4, Gemini, Grok, Cursor, Copilot, and local models.
- Clear scoring — Six key dimensions scored with concrete explanations and confidence levels.
- Prioritized fixes — Know exactly what to change first, with “why it matters” and “how to fix” guidance.
- Portability insights — Understand which parts will break when moving the skill between agents.
- Actionable report — Get a clean, copy-paste ready report you can use immediately.
Perfect for
- Anyone who writes skills or complex prompts across multiple agents
- People evaluating skills they found in marketplaces or shared by others
- Teams standardizing prompt/skill quality
- Creators who want to improve their skills before publishing
Zero friction. Paste a skill or prompt and get a clear diagnosis in seconds — completely free, no account required.
- Static analysis only: cannot test runtime execution.
- Quality depends on the depth of the provided prompt.
- Fixes require manual implementation.
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
31 uses
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 20 days ago
- Free with an Agensi account
Permissions
File Scopes
This is a lightweight diagnostic tool. It reads the material you provide and writes a report. It never executes or modifies the skill being analyzed. It works equally well when you simply paste text into any chat interface.