Citevana API and MCP server

REST API

Every response is JSON. Successful responses have "ok": true. The free tools return a plain-language summary and a to-do list ranked by importance, with time estimates.

Example request

curl -s https://citevana.com/api/v1/tools/ai-crawlers \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com"}'

curl -s https://citevana.com/api/v1/sites \
  -H "Authorization: Bearer $CITEVANA_API_KEY"

Errors

All errors use the same JSON shape, shown below. Error codes include invalid_request, invalid_url, blocked_host, fetch_failed, unauthorized, insufficient_scope, not_found, rate_limited and quota_exceeded.

{
  "ok": false,
  "error": {
    "code": "quota_exceeded",
    "message": "Daily api_calls quota reached."
  },
  "requestId": "req_…"
}

Limits and headers

Each metered response includes X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset (Unix seconds), describing whichever allowance the call was counted against. A 429 response also includes Retry-After. Keyless calls count against the shared free-tool allowance for your internet connection; keyed calls count against your workspace's daily api_calls allowance, and each key can have a lower daily cap of its own. A call that returns no result, because the site could not be reached or its bot protection answered instead, is not counted at all.

Browsers can call the free tool endpoints from any site without a key. Endpoints that need a key do not allow cross-origin browser requests, so call them from your server.

Add ?locale=es, pt-BR, de, fr, ja, id, hi, tr, vi or ar to a tool endpoint to get the summary in that language.

Source, checked September 16, 2026: https://spec.openapis.org/oas/v3.1.1.html

MCP server

Connect any MCP client that supports remote servers to https://citevana.com/api/mcp. It uses the Streamable HTTP transport, answers in JSON and keeps no sessions.

• check_ai_crawlers — Which AI search, assistant and training crawlers a site allows. Free.
• audit_page — On-page SEO and AI-readiness audit of one URL. Free.
• generate_llms_txt — Builds an llms.txt file from a site name, summary and links. Free; nothing is fetched.
• get_site_audit_summary — Latest audit for one of your sites. Needs an API key with read:audits.

The server supports protocol revision 2026-07-28 (per-request metadata and server/discover). It also accepts the initialize handshake of revisions 2025-11-25, 2025-06-18 and 2025-03-26.

The three free tools work without authentication, sharing the keyless allowance of 3 checks a day per internet connection. For get_site_audit_summary, your client must send your API key as an Authorization: Bearer header, which also raises that daily allowance. Citevana does not offer OAuth sign-in for MCP yet.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "check_ai_crawlers",
    "arguments": {
      "url": "https://example.com"
    },
    "_meta": {
      "io.modelcontextprotocol/protocolVersion": "2026-07-28",
      "io.modelcontextprotocol/clientCapabilities": {}
    }
  }
}

Source, checked September 16, 2026: https://modelcontextprotocol.io/specification/2026-07-28/basic/transports/streamable-http

Other MCP clients and your own agents

If your client lets you set HTTP headers for a remote server, add Authorization: Bearer followed by your key to unlock get_site_audit_summary. A call with a key counts against the same daily api_calls allowance as the REST API; without one it counts against the keyless free-tool allowance.

Chrome extension

The Citevana Chrome extension runs the AI-crawler check and the page audit on the tab you are viewing through this API. It needs no account and uses the keyless allowance; an optional key raises that limit and shows your own site audits. It is being prepared for the Chrome Web Store.

It asks only for activeTab and storage permissions, plus access to the Citevana API. It sends the page address only when you click a check, and never sends the query string.

Source, checked September 16, 2026: https://developer.chrome.com/docs/extensions/develop/concepts/activeTab

Safety and privacy

• Every tool is read-only. Nothing is ever changed on your website.
• Text copied from fetched pages is marked as untrusted data, so AI assistants should never follow instructions found in it.
• We fetch only public pages, as CitevanaBot, and we block private and reserved network addresses.
• We store only a SHA-256 hash of each API key. The key is shown once, and you can revoke it at any time. Creating and revoking keys is recorded in the audit log.

CitevanaBot

Frequently asked questions

Sources

• https://modelcontextprotocol.io/specification/2026-07-28/basic/transports/streamable-http
• https://modelcontextprotocol.io/specification/2026-07-28/basic/versioning
• https://developers.openai.com/apps-sdk/deploy/connect-chatgpt
• https://developers.openai.com/apps-sdk/build/auth
• https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp
• https://spec.openapis.org/oas/v3.1.1.html
• https://developer.chrome.com/docs/extensions/develop/concepts/activeTab
• https://developer.chrome.com/docs/webstore/program-policies/policies