# Legible

## What This Site Does
Legible scans websites and scores them 0–100 on how legible they are to AI agents — the Legibility Score. It identifies structural issues that prevent AI agents from reading, understanding, and navigating a site, and generates the fixes as finished artifacts.

## Primary Function
Enter any URL into the input field and select "Scan" to receive:
- A Legibility Score (0–100) with a grade (Agent-Ready, Mostly Legible, Needs Work, Agent-Hostile, or Agent-Blocked)
- A severity-ranked list of issues (Critical, High, Medium), each with a fix recommendation
- A downloadable agents.md file for the scanned site
- Machine-readable outputs: Policy Card (JSON), JSON-LD Actions, MCP Scaffold
- Agent Vision: side-by-side human vs. agent perception preview

## For Agents
- The scan is also exposed as an MCP tool. Discovery: `/.well-known/mcp.json`
- A score badge for any scanned domain is available at `/badge/{domain}`

## Interaction Guidance

### Allowed
- Browse and read all public pages
- Submit a URL via the input field to trigger a scan (the human user must confirm authorization)
- Download generated outputs (agents.md, Policy Card, JSON-LD, MCP Scaffold)
- Toggle dark/light theme

### Not Allowed
- No user accounts or authentication required or available
- No payment or personal data collection
- No bulk/automated scanning beyond normal usage

## Rate Limits
- Please wait for one scan to complete before initiating another
- Maximum 10 requests per minute

## Key UI Elements
- URL input: `#url-input` — accepts any https:// or http:// URL
- Scan button: triggers the scan
- Results appear in place of the hero after scanning completes
- Download buttons available after scan completes

## Contact
Built by Eddie Russell — eddieinsf.com