Your AI data, programmable, and AI-native
CiteTrack keeps everything on your own site and opens it up two ways: a clean REST API, hooks and WP-CLI for code and automation, and a built-in MCP server so you can plug your own Claude or ChatGPT into CiteTrack and just ask.
See the WordPress evidence behind the API and MCP tools
Watch the product workflow, then inspect current visibility, source and task records before connecting your own client.
Use CiteTrack as a connector for your own AI
CiteTrack ships a built-in MCP server (Model Context Protocol) at /wp-json/citetrack/v1/mcp. Create a named, expiring connection, choose its read, task, diagnostic and paid-run permissions, then connect a compatible assistant or automation client. Ask for comparable visibility movement, inspect saved work or check scan progress without giving every client permission to spend provider credit. CiteTrack does not copy the MCP dataset to a separate CiteTrack analytics cloud, but data returned through MCP is sent to the client you authorize and is then subject to that client's privacy terms.
Three ways to build
MCP connector
A built-in MCP server exposes visibility, citation, reputation, crawler, task and diagnostic tools to compatible assistants and automation clients through named, revocable connections.
REST API
Read crawls, citations, share of voice, reputation, page scores and alerts as JSON, and trigger runs or exports, under the citetrack/v1 namespace, authed with an Application Password.
Hooks & WP-CLI
Fire your own code when a citation is recorded or an alert trips, and run checks, exports and maintenance from the command line or cron, ideal for managed fleets.
The MCP tools, at a glance
Each maps to a tested REST endpoint, so there's one source of truth.
get_snapshot, health, citations, share of voice, reputation in one call.list_citations/share_of_voice, per-prompt results vs competitors.indexed_not_cited, a bounded compatibility observation only; it excludes unmeasured and technical URLs and never treats crawling as proof of indexing or a task.reputation_changes, recent drops, sentiment shifts, flagged claims.recent_crawls/list_prompts, raw activity and your tracked set.list_tasks/get_scan_status, saved work and bounded diagnostics without starting new work.run_citations, queue one selected prompt only after explicit spend confirmation and with a dedicated run permission.
{
"mcpServers": {
"citetrack": {
"type": "http",
"url": "https://yoursite.com/wp-json/citetrack/v1/mcp",
"headers": { "Authorization": "Bearer YOUR_TOKEN" }
}
}
}add_action(
'citetrack_alert_triggered',
function ( $alert ) {
wp_remote_post( SLACK_WEBHOOK, array(
'body' => wp_json_encode( array(
'text' => $alert['message'],
) ),
) );
}
);Hooks you'll actually use
Tie CiteTrack into the rest of your stack with standard WordPress hooks.
citetrack_citation_recorded, a prompt/engine result was written.citetrack_score_changed, a page's citability/AEO score moved.citetrack_alert_triggered, a rule fired (route it anywhere).citetrack_bot_signatures, extend the crawler catalog with your own bots.
Built for ownership
Because every metric lives in your own WordPress database, none of this is a window into someone else's cloud, it's direct access to your data. Warehouse it, build a custom client dashboard, wire alerts into Slack and PagerDuty, or let your own AI assistant query it through MCP. Agencies roll site-level data into multi-client views without breaking the promise that each client's data stays on their own site.
Authentication & safety
REST endpoints use standard WordPress auth with capability checks (an Application Password over HTTPS). MCP uses named, expiring and individually revocable connections with separate read, tasks, diagnostics and paid-run permissions. A paid citation check requires one active prompt ID plus explicit spend confirmation; queue acceptance does not claim the check completed.
Developer API & MCP FAQ
Can I connect CiteTrack to Claude or ChatGPT?
Yes. CiteTrack includes a built-in MCP server at /wp-json/citetrack/v1/mcp. Create a named connection on the API & MCP screen, choose its permissions and expiry, then add it as a remote connector in a compatible client.
What MCP tools are available?
get_snapshot, list_citations, share_of_voice, indexed_not_cited, reputation_changes, crawl_health, recent_crawls, list_prompts, list_tasks, get_scan_status and run_citations. The legacy indexed_not_cited tool returns bounded observations, not optimization tasks, and crawling does not prove indexing.
Can an MCP client spend API credit automatically?
Not through a normal read connection. Paid citation checks require the separate run permission, one active prompt ID and explicit confirm_spend approval. Requests are rate-limited, audited and reported as queued rather than complete.
Do I have to use MCP?
No. The same data is available as plain REST under citetrack/v1, authenticated with a WordPress Application Password, perfect for n8n, Make, scripts or your own dashboard.
What can the REST API return?
Crawl-log data, citation results (status, position, winning URL, share of voice), reputation events, per-page citability/AEO scores, alerts and report snapshots, all JSON.
Is there WP-CLI and hooks support?
Yes, action/filter hooks fire when citations are recorded, scores change and alerts trigger (plus a filter to extend the bot catalog), and WP-CLI runs checks, exports and maintenance from the command line or cron.
Does any of this send data to CiteTrack?
The REST API and MCP server read from your WordPress database and do not copy the response into a separate CiteTrack analytics cloud. An authorized MCP, automation or API client receives the data it requests, so review that client privacy and retention terms.
Your AI data, on tap
REST, hooks, WP-CLI, and a built-in MCP server so your own AI can query it. All on your site.
Explore: Features · Crawl Academy · Glossary · Compare · Pricing