MCP server

Engadin exposes the workspace as a Model Context Protocol server, so an AI assistant of yours — Claude, an MCP Inspector, your own agent — can read conversations, leads, analytics and gaps, and trigger a crawl, without you building an integration for each of those.

Endpoint — https://engadin.ai/api/mcp (streamable HTTP, stateless)

Connecting

Two ways, both configured on Developers → MCP.

OAuth

Full OAuth 2.1 with dynamic client registration, PKCE and refresh tokens. An MCP host that speaks OAuth — claude.ai, the MCP Inspector — needs only the endpoint URL; it discovers the rest:

https://engadin.ai/.well-known/oauth-protected-resource
https://engadin.ai/.well-known/oauth-authorization-server

You approve the connection once, in the browser, signed in as yourself. The resulting access is scoped to your workspace and your role.

Bearer token

For hosts that do not do OAuth, mint a token on the MCP page and send it:

Authorization: Bearer <mcp token>

Caution

MCP tokens and REST API keys are different kinds and are not interchangeable.

If you registered against an older hostname

OAuth clients are bound to the issuer they registered with. Older platform hostnames keep serving /api/mcp, /oauth/* and the discovery documents permanently and identically, so an existing registration keeps working — you do not need to re-register to keep it alive. New connections should use engadin.ai.

Tools

Everything is scoped to the caller’s workspace, and every call is logged.

Tool	Does
list_agents	Your agents with id, name, website and status.
get_stats	Counts: conversations, messages, leads, indexed pages, queue depth, average sentiment.
get_analytics	Conversations and leads by day, country and language; conversion rate; average duration.
list_conversations	Recent conversations with agent, country, language, message count, sentiment, topics.
get_conversation	One full transcript, with retrieval scores, ratings and any lead.
search_conversations	Substring search across visitor and assistant messages.
list_leads	Captured leads, newest first.
content_gaps	Questions answered on weak evidence — candidates for new pages.
content_gap_progress	Which gaps have gone quiet since a date, and which have not.
content_gap_briefing	A writing brief for one gap.
crawler_status	Page counts, queue depth, pages with errors, last crawl times.
trigger_crawl	Re-seed from the sitemap and process one batch now.
list_action_runs	Recent custom action runs and their outcomes.

Most tools take an optional agent id; without one they cover every agent the caller can see.

Configuring a client

For an MCP host that reads a JSON config:

{
  "mcpServers": {
    "engadin": {
      "url": "https://engadin.ai/api/mcp"
    }
  }
}

Hosts that support OAuth will prompt you to authorize on first use. Hosts that do not will need the bearer token added the way that host expects.

What it is good for

• “Has anyone asked about X?” — search_conversations answers in seconds what scrolling the inbox does not.
• Weekly content planning — content_gaps plus content_gap_briefing produces briefs your writers can work from directly.
• After a launch — trigger_crawl, then crawler_status, then ask the agent the questions the new pages should answer.

What it will not do

There is no tool that changes settings, edits agents, manages members or deletes anything. The MCP surface is read-plus-crawl on purpose: it is reachable by a language model acting on instructions from text it read, and that is not a boundary to put configuration behind.