§ /mcp · streamable-http server

Audit any URL from Claude Desktop in three lines.

A Model Context Protocol server at /api/mcp exposes two tools (audit_url and get_quota): to MCP-capable clients (Claude Desktop, Cursor, ChatGPT desktop, Claude API connectors). Same engine as the CLI and web app. Bearer-auth with the same key. No separate billing.

§ connect (claude desktop)

Add the server to your claude_desktop_config.json:

{
  "mcpServers": {
    "nyman-ai-visibility": {
      "type": "streamable-http",
      "url": "https://app.nyman.media/api/mcp",
      "headers": {
        "Authorization": "Bearer nmav_live_..."
      }
    }
  }
}

Get a key from Account settings → API access on the web app (Premium subscription required). Restart Claude Desktop and ask: "Run an AI visibility audit on example.com."

§ tools exposed

audit_url

Run a deterministic AI-readiness audit on a URL. Returns score, hard_block status, section breakdowns, and the canonical share_url for the result. Counts against your daily audit quota.

get_quota

Introspect the current key: tier, remaining audits, monthly caps. Non-metered, never consumes quota.

§ why

Almost no other tool in this category ships an MCP server. As MCP clients grow into the default surface for AI agents to act on the web, having the audit engine wired into Claude / Cursor / ChatGPT means the agent can audit, fix, and re-audit a site without leaving the chat. That's a real difference, not a checkbox.

§ same engine, every surface

clinmav audit https://example.com
restPOST /api/v1/audits
mcptools/call audit_url
webpaste a URL at nyman.media

full docs →cli →rest api →pricing →