Choosing an Integration Surface

TimeBack exposes three integration surfaces. They are not alternatives for the same job: REST is the contract you build a product on, MCP is how an AI assistant explores, and webhooks are how TimeBack tells you something happened. Most integrations use more than one.

Which Surface for Which Job

If you want to Use Why
Read one student's insights in your product REST (GET /insights/1.0/users/{userId}) Stable contract, pagination, filtering, your own credentials
Discover which insight types exist REST (GET /insights/1.0/types) The catalog is not reachable from any MCP server today
Explore data ad hoc, or let an AI assistant answer questions Analytics MCP SQL over curated views; 500-row and 25-second limits
Let an AI assistant look up a student's sessions conversationally Platform MCP Scoped to session insights for the signed-in user
Register an app, manage credentials or webhooks Developer-platform MCP, or the REST applications API Purpose-built tools; two of its tools are not yet available
Be told when a session ends or an insight is raised Webhooks Push, no polling

The Three MCP Servers

Server URL Auth For
Platform MCP https://platform.timeback.com/mcp OAuth 2.0 with dynamic client registration, Google sign-in Assistants working across platform data
Analytics MCP https://platform.timeback.com/mcps/analytics OAuth, org-scoped Analytical questions over curated views
Developer-platform MCP https://platform.timeback.com/mcps/developer-platform Same OAuth as the platform MCP Developers integrating an app

When REST Is Still the Right Answer

MCP exists alongside REST, not instead of it:

  • MCP tool surfaces change as we improve them and are not a versioned contract.
  • The platform MCP's data reads are GET-only and scoped to the signed-in user. It can reach session insights (/insights/1.0/sessions and /insights/1.0/sessions/{sessionId}) and nothing else under /insights/ — including the type catalog.
  • The analytics MCP caps results at 500 rows and 25 seconds, and answers from curated analytics views rather than the live Insights API.
  • Anything you ship to your own users belongs on REST.

See also Level 4: Insights API, Analytics, and Webhooks.