AI Readiness Audit Skill
Audit any website for AI agent readiness using the Inlay API. Checks 11 categories including llms.txt, MCP servers, structured data, semantic HTML, meta quality, and more.
Quick Start
Ask the user for a URL, then run the audit:
curl -s -X POST https://www.inlay.dev/api/audit \
-H 'Content-Type: application/json' \
-d '{"url":"TARGET_URL"}'
Or use the wrapper script:
bash scripts/audit.sh "https://example.com"
Workflow
Step 1: Get the Target URL
Ask the user which website to audit. Accept any valid URL.
Step 2: Run the Audit
curl -s -X POST https://www.inlay.dev/api/audit \
-H 'Content-Type: application/json' \
-d '{"url":"TARGET_URL"}'
The API returns a JSON response with:
score— overall score (0-100)grade— letter gradecategories— per-category scores and findingsrecommendations— actionable fixes sorted by priorityboostScore— projected score after applying Inlay Boost (if available)
Step 3: Present the Report
Format the results as a clear report. See examples/sample-report.md for the expected format.
Report structure:
- Header — Site URL, overall score, letter grade
- Grade Scale — A+ (90-100), A (80-89), B (70-79), C (60-69), D (40-59), F (0-39)
- Category Breakdown — Table with each category's score and status
- Top Issues — Negative findings that hurt the score
- Recommendations — Actionable fixes sorted by impact (high → low)
- Inlay Boost — Projected score if Inlay Boost data is available
Step 4: Offer to Fix Issues
After presenting the report, offer to fix issues automatically:
- llms.txt missing → Use the
setup-llms-txtskill to create one - No MCP server → Use the
setup-mcp-serverskill to set one up - Missing structured data → Generate JSON-LD schema markup
- Poor meta tags → Rewrite title/description for AI discoverability
- Missing robots.txt directives → Add AI bot permissions
- No sitemap → Generate or update sitemap.xml
For each fixable issue, explain what it is, why it matters for AI agents, and offer to implement the fix in the user's codebase.
Categories Reference
See references/scoring.md for full details on all 11 audit categories:
| Category | Weight | What It Checks |
|---|---|---|
| llms.txt | High | Presence and quality of llms.txt / llms-full.txt |
| MCP Server | High | MCP endpoint availability and tool quality |
| Structured Data | High | JSON-LD, schema.org markup |
| Meta Quality | Medium | Title, description, Open Graph tags |
| Semantic HTML | Medium | Proper heading hierarchy, landmarks, ARIA |
| Robots & Crawling | Medium | robots.txt AI bot permissions, sitemap |
| Performance | Medium | Load time, Core Web Vitals signals |
| Security | Low | HTTPS, headers, content security |
| Accessibility | Low | Basic a11y signals |
| Content Quality | Medium | Readability, structure, depth |
| AI Signals | High | Overall AI-specific discoverability markers |
Common Fixes
See references/fixes.md for detailed fix instructions for each category.
Tips
- Run audits on both the homepage and key inner pages
- Compare scores before/after implementing fixes
- Focus on high-weight categories first for maximum impact
- The Inlay Boost projected score shows the potential improvement from using Inlay's tools