Legible, inside your AI client
Legible is a Model Context Protocol server. Connect it to Claude, Cursor, or any MCP-capable assistant and score how findable, citable, and buyable any brand is to AI — without leaving the chat. A company built to make you legible to agents should be callable by them.
Endpoint
https://legibleindex.com/api/mcpStreamable HTTP · no API key required.
Tools
- score_brand(url)Free
The Legible Readiness Index (0–100) across 8 dimensions, a band, and prioritized gaps to fix.
- get_fix_pack(url)Free preview
A guided preview of what to fix (structured data, llms.txt, robots, ai-facts, FAQ, hreflang). The full ready-to-paste pack and the deep modules live in the Legible web app.
Connect
Most MCP clients accept a remote URL directly. Add Legible to your client config:
{
"mcpServers": {
"legible": { "url": "https://legibleindex.com/api/mcp" }
}
}- · Claude Desktop — Settings → Developer → Edit Config (
claude_desktop_config.json), then restart. - · Cursor — Settings → MCP → Add server, paste the URL.
- · Other clients — use the URL anywhere a remote MCP server is accepted.
If your client only supports local (stdio) servers, bridge it:
npx mcp-remote https://legibleindex.com/api/mcpOnce connected, ask your assistant:
“Use Legible to score stripe.com — what should they fix first?”
What's included
- score_brand — full LRI + gaps
- get_fix_pack — guided fix preview
- Full Fix Pack (every copy-paste fix)
- Deep Audit · Verification · Share-of-Voice
The MCP tools are free and need no account. The full Fix Pack and the deep modules are part of Legible Pro (CHF 49/mo), available on demand in the web app.
Limits
- · Rate limit: ~40 requests/min and 300/day per IP.
- · Each scan fetches a public URL (private/internal hosts are blocked).
- · No authentication in v1 — anonymous and free.
Prefer the web app? Run a free report →