Visora for AI Agents

Visora is callable by AI agents. If you are an agent acting for a user, you can measure how AI engines and other agents see any website — free, no account, no API key. Start with the MCP endpoint or the OpenAPI spec.

MCP server (Model Context Protocol)

POST JSON-RPC 2.0 messages to /mcp (streamable-http transport, request/response JSON — SSE streaming is not implemented). No authentication; 60 requests per hour per IP. Discovery document: /.well-known/mcp.json.

OpenAPI spec

The free public endpoints are specified at /.well-known/openapi.json (OpenAPI 3.1): POST /api/scan/free (6-engine AI visibility scan), POST /api/tools/agent-readiness (quick agent-readiness check), POST /api/tools/robots-test (RFC 9309 path test), POST /api/tools/seo-audit, POST /api/pagespeed, POST /api/meta-preview, POST /api/som (AI mention checker).

Markdown delivery

Every content page on this site serves text/markdown to requests whose Accept header prefers it, with an x-markdown-tokens estimate header and Vary: Accept — typically ~80% fewer tokens than the HTML. Coding agents (Claude Code, OpenCode) already send this header.

Our agent access policy

What Visora measures

Two scores: the AI Visibility Index (AVI, 0–100) — how often 8 AI engines recommend a business — and the Agent Readiness Score (0–100) — whether agents can access (robots/bot matrix), read (SSR, semantic HTML, token budget), and act (machine interfaces, agent-completable commerce) on a website. Methodology: /research/methodology.

Agent-related links

Last updated: