# Legible

> Legible audits how well AI agents can read, understand, and navigate a website — and generates the fixes. Scan any site to get a Legibility Score (0–100), a grade, a severity-ranked issue list with fix recommendations, and four generated artifacts: agents.md, JSON-LD Actions, an agent policy card, and an MCP server scaffold.

Legible reads pages the way agents do — structure, labels, and meaning — and runs 20 named checks across interaction clarity, structure, and machine-readable meaning. The score is explained, never a black box. Grades: Agent-Ready (90+), Mostly Legible, Needs Work, Agent-Hostile, and Agent-Blocked for sites that turn agents away.

## Key resources

- [Home / scanner](https://legible.page/): scan any URL from the input on the landing page
- [agents.md](https://legible.page/agents.md): how agents should use this site
- [MCP server card](https://legible.page/.well-known/mcp.json): call the scan as an MCP tool
- [Score badge](https://legible.page/badge/legible.page): embeddable SVG Legibility Score badge for any scanned domain

## Notes

- Scans respect robots.txt and require the human user to confirm they are authorized to audit the target site.
- Built by Eddie Russell.