Skip to main content

Documentation Index

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

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

Custory includes an MCP server so external AI clients can work with workspace context directly.

​
What MCP access is for

Use MCP when you want tools such as Cursor, Claude, or other compatible agents to:
  • Read workspace context
  • Work with journeys and related data
  • Participate in broader product or engineering workflows without leaving the AI client

​
Workspace-scoped keys

MCP access uses workspace-scoped API keys. That matters because access stays tied to the right workspace instead of using a broad credential with unclear scope. From Manage Integrations, you can:
  • Create a new MCP key
  • Set an expiry window
  • Copy setup snippets
  • Revoke old keys
You can only view the full token at creation time, so treat it like a secret.

​
Connection details

Custory provides setup guidance directly in the product, including ready-to-paste configuration for common clients and a quick test request. The MCP endpoint is exposed as a dedicated route and expects authenticated access using the workspace-scoped token.

​
Operational behavior

Custory tracks MCP connections and logs MCP tool usage so teams have a record of how external AI clients are interacting with the workspace.

​
When to use MCP instead of in-product AI

Use in-product AI when the work belongs naturally inside Custory and benefits from the product UI. Use MCP when the team already works from an external AI client and wants Custory to participate directly in that broader workflow.