Connecting the MCP server

The MCP server lets an AI assistant — Claude, or any MCP-compatible client — work with your FeedbackRobot data directly.

Why you’d want this

The Assistant inside FeedbackRobot answers questions about your feedback. The MCP server does the same thing from wherever you already work, alongside your other context.

That matters because the interesting questions usually aren’t only about feedback. “Draft a response to this complaint using our refund policy” needs the complaint and the policy. “Compare our review sentiment against last quarter’s sales” needs both sets of numbers. An assistant with access to your feedback and your other tools can answer things neither could alone.

What it can do

Once connected, an AI client can:

• List and read your reviews, filtered by platform, sentiment, rating or date
• List surveys and read responses
• Look up a customer and see their history with you
• Check sentiment distribution and trends
• Review open and closed resolutions and their success rates
• Draft and send surveys
• Trigger and follow up resolutions

Read operations are safe to explore with. Anything that sends or triggers affects real customers — be deliberate about which client you grant that to.

Connecting

You’ll need your FeedbackRobot credentials and an MCP-compatible client. In your client’s connector settings, add FeedbackRobot and authorise it when prompted.

For Claude, that’s done through connector settings — the client walks you through authorisation.

Security

• The connection uses your account’s permissions. It can’t see more than you can.
• Grant it only to clients you trust. An AI client with send permissions can email your customers.
• Revoke access from your workspace settings if you stop using it.

→ Security and acceptable use

Useful things to try

• “Which customers left negative reviews this month and what did they say?”
• “Draft replies to the three open resolutions, matching our usual tone.”
• “What’s our resolution success rate, and which failures have anything in common?”

API access

For programmatic integration rather than AI-assisted work, see using the API.