MCP server: connect your AI assistant

Encited ships a remote MCP (Model Context Protocol) server, so any MCP-capable AI assistant can work with your Encited data directly: pull crawl analytics, inspect cached snapshots, read AI-mention results, and — with your permission — start SEO audits and content workflows.

Connect your assistant#

The server lives at:

https://encited.com/api/mcp

Authentication is OAuth — your assistant opens a browser window where you sign in to Encited and approve the connection. No API keys to copy.

ChatGPT: Settings → Apps & Connectors → Advanced settings → enable Developer mode (required for custom connectors), then Create connector: give it a name, paste the URL above, choose OAuth authentication, and complete the sign-in when the browser opens.

Claude (claude.ai): Settings → Connectors → Add custom connector, paste the URL above, and complete the sign-in when the browser opens.

Claude Code:

claude mcp add --transport http encited https://encited.com/api/mcp

Then run /mcp in a session to trigger the sign-in.

Codex (CLI): add the server to ~/.codex/config.toml:

[mcp_servers.encited]
url = "https://encited.com/api/mcp"

Then run codex mcp login encited to trigger the sign-in.

Cursor: add the server to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "encited": {
      "url": "https://encited.com/api/mcp"
    }
  }
}

Then complete the sign-in when Cursor prompts you.

The consent screen#

When you approve a connection you choose two things:

• Which account the assistant acts on — your personal account or a team you belong to. One connection is tied to one account; connect again to use another.
• What it may do:

Write access is only offered when your role in the selected account allows it (editor or above). Viewers and read-only members get read-only connections.

What your assistant can use, by plan#

Everything below inherits your existing plan limits — audit page quotas, workflow credits, daily caps — exactly as in the dashboard.

Tools outside your tier simply don't appear in the assistant's tool list. Team members only see the domains they have access to — domain-scoped members included.

Tool reference#

Tools marked write require write permission on the connection.

General

• list_domains — the domains in the connected account; the starting point for every other tool.

Snapshot cache

• list_snapshots — cached prerendered pages for a domain, with age and size.
• get_snapshot — one cached page's metadata and SEO analysis, optionally with the rendered HTML.

Crawl analytics

• query_crawl_logs — recent bot and crawler visits, filterable by provider, path, status, outcome, and category.
• get_crawl_stats — aggregate crawl activity: top crawlers, pages crawled, crawl-budget savings, per-day breakdown.

Google Search Console

• get_gsc_connection_status — whether Search Console is linked for a domain and whether ownership is verified.
• get_gsc_keywords — top search queries with clicks, impressions, click-through rate, and average position.
• get_gsc_page_analytics — impressions and clicks per page, best performers first.
• get_gsc_site_health — totals for the period: clicks, impressions, average position, pages with impressions.
• inspect_gsc_url — one URL's index status straight from Google, including why it isn't indexed.

SEO audits

• list_seo_audit_runs — audit history with health scores and issue counts.
• get_seo_audit_run — one audit's per-page results (issues, head fields, content metrics), worst pages first.
• get_seo_audit_page — one audited page's full record: head fields, social tags, response metadata, and internal links.
• get_seo_audit_issues — current issues across the domain, grouped by type.
• start_seo_audit (write) — start a crawl: follow links, audit selected URLs, or discover from the sitemap.

Index Rush — Essential and above

• get_index_rush_pages — sitemap pages joined with Google index status and inbound-link coverage.
• get_index_rush_inbound_links — which indexed pages link to a not-yet-indexed target.

Content Workflows — Essential and above

• list_workflows / get_workflow — the workflow catalog and each workflow's inputs and credit cost.
• get_workflow_credits — your credit balance for the current cycle.
• start_workflow (write) — run a workflow; costs credits.
• list_workflow_runs / get_workflow_run / get_workflow_result — run history, per-step progress, and the finished output.

AI Mentions

• list_ai_prompts — the prompts monitored for a domain.
• list_ai_prompt_runs / get_ai_prompt_run — per-provider run results: mention position, sentiment, citations, analysis.
• list_competitors — brand vs competitor visibility across AI answers.
• list_citations — the domains AI providers cite, with counts and classification.
• list_fanout_queries — the web searches providers ran while answering (Pro and above).

Managing connections#

Every connected assistant appears in Settings → Connections, with its name, permissions, and connection date. Revoking a connection cuts the assistant off immediately — its next request fails and it must re-authenticate.

On Business and Enterprise plans, connection approvals and revocations are recorded in the audit log under the MCP category.

Good to know#

• Changed your plan? The assistant's tool list updates the next time it reconnects — in Claude Code, run /mcp and reconnect; in claude.ai, toggle the connector.
• Your assistant knows its way around. The server ships built-in usage guidance: when it connects, your assistant learns which tool fits which task (from "is Google crawling my site?" to "check Search Console for growth opportunities") without you having to explain the toolkit.
• Write tools missing? Your connection is read-only — either you didn't grant write access, or your role in that account doesn't allow it. Re-connect and approve write access from an editor-or-above account.
• Left a team, or the plan lapsed? Connections stop working on the very next request — no waiting for tokens to expire.
• Rate limits. Each account gets 120 MCP requests per minute — far more than an assistant needs in normal use. Over the limit, requests return a rate_limited error with a Retry-After header; the assistant can simply wait and retry.
• Assistants never see raw API keys, internal identifiers, or other accounts' data; every request is re-checked against your current membership, role, and plan.