TeamRetro MCP
Coming soon
We're currently testing MCP support with a small group of customers ahead of general release. If you'd like early access, please contact support.
Connect TeamRetro to your preferred AI assistant
Bring TeamRetro's data into AI assistants like ChatGPT, Claude, Cursor, and others. Once connected, you can ask natural-language questions like "What were the top action items from last sprint's retro?" or "Summarise our team's last health check" — and the AI pulls the answer straight from TeamRetro.
This works through MCP (Model Context Protocol), an open standard for connecting AI assistants to your data.
Before you begin
- A TeamRetro account on a plan that includes API & MCP access
- An AI assistant that supports MCP (see the list below)
- Permission to grant connected-app access on your account (account admins can manage this for everyone — see Manage connected apps below)
Connect TeamRetro to your AI assistant
The TeamRetro MCP server lives at:
https://mcp.teamretro.com/mcp
You add this URL to your AI assistant. The first time you use it, your browser will open and ask you to sign in to TeamRetro and approve the connection — there's no API key to copy or paste.
One account per connection: each connection is linked to a single TeamRetro account. If you belong to more than one account, connect each one separately — most AI assistants let you add the same MCP server multiple times under different names (e.g. TeamRetro — Acme, TeamRetro — Acme2). You'll choose which TeamRetro account to use during each sign-in.
ChatGPT
Available on ChatGPT Pro, Business, Enterprise, and Edu plans.
- In ChatGPT, open Settings → Connectors (or Custom Connectors).
- Click Add connector and choose MCP as the type.
- Paste
https://mcp.teamretro.com/mcpas the server URL. -
Click Connect. Sign in to TeamRetro in the browser window and choose the account you want to connect.
Claude (claude.ai)
- In Claude, open Settings → Connectors.
- Click Add custom connector.
- Paste
https://mcp.teamretro.com/mcpas the server URL and give it a name (e.g. TeamRetro). -
Click Connect and approve the connection in the browser window that opens.
Claude Desktop / Claude Code
Add TeamRetro using the built-in MCP command:
claude mcp add --transport http teamretro https://mcp.teamretro.com/mcp
Then run /mcp inside Claude. You'll be prompted to Authenticate — click through, sign in, and choose the account you want to connect.
Cursor
In Settings → MCP, add a new server:
- Name: TeamRetro
- URL:
https://mcp.teamretro.com/mcp - Transport: Streamable HTTP
Cursor will open a browser window the first time it connects.
Other MCP clients
Any client that supports the Streamable HTTP transport and OAuth 2.1 will work — including Windsurf, Zed, and VS Code with GitHub Copilot. Use:
- Server URL:
https://mcp.teamretro.com/mcp - Authentication: OAuth (the client handles registration and consent automatically)
What you can ask
Once connected, try prompts like:
- "List my TeamRetro teams"
- "Summarise the last retrospective for the Platform team"
- "What overdue actions does the Marketing team have?"
- "How has our latest health check trended?"
Manage connected apps
You can see which AI assistants are connected and revoke any of them at any time.
- Your connections: Profile → Connected apps
- Account-wide (admins): Account -> Settings → Connected apps to see and revoke connections across all users
Revoking immediately invalidates the AI assistant's access. To reconnect, just run the connection step in your assistant again.
What data is shared?
When you approve a connection, you grant a specific list of permissions ("scopes") — for example, read your teams and read your meetings. Your AI assistant only sees the data those scopes allow, and only for the TeamRetro account you selected during setup.
TeamRetro never shares your data with the AI provider unless you ask the assistant a question that requires it.
Troubleshooting
- The assistant says "fetch failed" or "can't connect": Double-check the URL is exactly
https://mcp.teamretro.com/mcp. Some clients require the trailing/mcppath. - I'm signed in to TeamRetro but the consent screen keeps redirecting to login: Sign out and back in to your TeamRetro account, then retry the connection.
- My AI assistant isn't on the list above: Any MCP client that supports Streamable HTTP + OAuth 2.1 should work. Check your assistant's MCP documentation for how to add an HTTP server.
Looking for the developer API?
Building your own integration directly against TeamRetro? See the TeamRetro API documentation for REST endpoints and authentication options including long-lived API keys.