The Retently MCP lets you work with your Retently data directly from Claude Code. Once connected, you can ask Claude to pull responses, build reports, look up customers, and run actions in your account from the terminal.
This article covers adding the MCP to Claude Code. If you use Claude Cowork instead, see Connect the Retently MCP to Claude Cowork. For what you can do once connected, see What the Retently MCP Can Do.
The MCP is still being actively improved. If something is missing or could work better, send us your feedback, ideally with the exact prompt you used, so we can optimize for that use case.
Before You Start
You need to be logged in to the Retently account you want to connect. Claude links to whichever account is active in your browser during the authentication step.
Set your AI access permissions
Before connecting, open your AI Connections page in Retently. Under Customer Personal Identifiable Information Access, you will find two toggles. By default, both are switched off. They control whether Claude can read personal identifiable information (PII) and whether it can manage personal data.
What each state means:
On: Claude can read contact and customer properties, use filters, attribute survey responses to the customers who left them, and perform write actions such as sending surveys, tagging, and assigning.
Off (default): Claude can pull aggregate, quantitative data only. It cannot read contact properties, use filters, attribute responses to specific customers, or perform any write actions.
Most of the MCP's most useful capabilities depend on these toggles. Filters, contact properties, customer attribution, and every write action are tied to the PII read and manage settings. With both toggles off, you are limited to anonymous, quantitative data. Unless you specifically need to keep your data anonymous, we recommend enabling both.
If you want to let someone work with quantitative data without exposing PII, consider creating a separate, tightly scoped access token for that purpose instead of leaving the toggles off.
Set these toggles before you connect. The connection captures your access level when you authenticate, so if you change a toggle afterward (for example, you connected with PII access off and now want to switch it on), reconnect and re-authenticate the Retently MCP for the change to take effect.
Step 1: Add the MCP Server
From your terminal, add the Retently MCP server to Claude Code:
claude mcp add --transport http retently https://mcp.retently.com/mcp
The MCP server URL is https://mcp.retently.com/mcp. The command registers it under the name retently.
You can also copy the ready-made command from the AI Connections page in your Retently account.
Step 2: Restart Claude Code
You need to restart Claude Code before the new connection becomes active. A session loads its available MCP tools when it first starts, so a server you add mid-session is not picked up until you restart.
After restarting, Retently appears in your list of MCP servers with a status indicating it needs authentication.
Step 3: Authenticate
Connecting the server does not authenticate it automatically. You have to authenticate it yourself after the restart:
Run the
/mcpcommand in Claude Code.Find the Retently server in the list and select it.
Choose Re-authenticate (or Authenticate).
Claude Code redirects you to your browser, where your Retently account is logged in.
A permissions screen appears showing what the connection will be allowed to do. This screen mirrors the two toggles from the AI Connections page. If you left them off in your account, you can enable them here to grant that access for this connection.
Click Allow access.
Once you return to the terminal, the Retently server shows as authenticated and ready to use.
If the connection ever drops or you need to change access, run /mcp again and re-authenticate the Retently server.
You Are Connected
Retently is now available in Claude Code. A good first step is to ask Claude to list everything it can do:
What can the Retently MCP do? List all the tools and give a short description of each.
From there, ask the kinds of questions you normally ask yourself when looking at your Retently data. See What the Retently MCP Can Do for categories of tools and example prompts.
Related