Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ontora.com/llms.txt

Use this file to discover all available pages before exploring further.

Ontora hosts a Model Context Protocol (MCP) server at https://api.ontora.com/mcp/. Once an MCP-aware client is configured with a workspace API key, the agent on the other end can list campaigns, read transcripts, run GraphRAG queries, manage webhooks, and more — without writing any glue code.

​
When to use the MCP server

Use caseBest fit
Letting an AI agent operate Ontora end-to-endMCP
Building a backend integration with predictable shapesREST API
Driving Ontora from terminal or CICLI
The MCP server and REST API expose the same data and obey the same scope rules — they’re two interfaces over the same backend.

​
Authentication

MCP requests authenticate with a workspace API key that has the mcp scope.
  1. Create an API key with the mcp scope (and any other scopes the client will need — typically interviews, exports, and webhooks).
  2. Configure your MCP client to send the key as a Bearer token.
  3. The workspace is derived from the key — no extra headers needed.
Authorization: Bearer ont_live_...
Because the workspace is bound to the key, an MCP client always operates inside a single workspace. To switch workspaces, swap the key.

​
What the agent can do

The MCP server exposes ~17 tools across these areas:
  • Campaigns — list, get, create, start, pause, add contacts, monitor progress
  • Transcripts — list conversations, fetch transcripts, export Markdown
  • Synthesis — get roadmap, get personas, run GraphRAG queries, export full report
  • Webhooks — list, add, remove subscriptions
See the full reference: MCP tools.

​
Next steps

Connect a client

Step-by-step setup for Claude Desktop, Cursor, and Windsurf.

Tool reference

Every MCP tool, with parameters and return shapes.