Skip to content
English
Go to remesh.ai
  • There are no suggestions because the search field is empty.

The Remesh MCP Server

This article is relevant to Live and Flex Conversations. You can find more information about our product offerings here.

Overview

The Remesh MCP server lets you connect your own LLM client (GPT, Claude, Co-Pilot, Gemini, etc) directly to your Remesh workspace, so you can browse, build, and analyze conversations from outside the Remesh UI. It is for researchers and analysts who want to work with their Remesh data inside the AI tools they already use.

The MCP server only exposes data and actions you already have access to in Remesh. Connecting a client does not grant new permissions.

Included in this Article

What is an MCP Server?

MCP (Model Context Protocol) is an open standard that lets an AI client talk to an external service through a defined set of tools. The Remesh MCP server publishes Remesh's conversation, setup, and analysis capabilities as MCP tools, so a connected client can call them on your behalf.

In practice, this means you can ask your AI client to do things like "list my conversations in this workspace," "build a Flex from this outline," or "what did participants disagree on most," and it will call Remesh directly and return the results.

Connect an AI Client

You connect a supported MCP client to your Remesh workspace and authenticate. Once connected, you select the workspace you want the client to work in.

Step

What happens

Add the Remesh server to your client

Your client registers the Remesh MCP server as a tool provider.

Authenticate

You sign in and authorize the client to act on your behalf.

Select a workspace

The client works within the workspace you choose; each client keeps its own workspace selection.

What You Can Do

After you connect a client, you can use the following types of Remesh tools:

  • Workspace and Conversation setup: View teams, folders, and Conversations. You can also open, create, duplicate, schedule, and edit Conversations and their sections.

  • Remy workflows: Use Remy to analyze data, create a discussion guide with AutoDG, or get help using the Remesh platform.

  • Analysis tools: Run focused analyses, including top responses by agreement, segment differences, sentiment analysis, and crosstabs.

Review Sources and Citations

Analysis responses include the same numbered citations you see in Remesh. Each citation points to a source used in the response. In clients that support linked citations, select a citation to open the source view in Remesh and review the supporting data.

Security and Access

  • Authentication: You must sign in and authorize each connection before using Remesh tools.

  • Existing permissions: The MCP server follows your current Remesh permissions. You can only access the workspaces, Conversations, and data available to your Remesh account.

  • Rate limits: Requests are limited by workspace and user to help protect platform performance.

  • Separate sessions: Each connected client keeps an independent workspace selection.

Frequently Asked Questions

Which AI clients can I use?
The Remesh MCP server can connect to supported MCP clients, including GPT, Claude, and Gemini. The setup and available features may vary by client.

Does connecting an AI client give it more access than I have?
No. The MCP server follows your existing Remesh permissions and cannot access Conversations or data that are not available to your account.

Do analysis responses include sources?
Yes. Analysis responses include numbered citations that point to the supporting source in Remesh. Linked citation behavior depends on the AI client you use.

Can I create and edit Conversations through an AI client?
Yes. Available setup tools can let you create, duplicate, schedule, and edit Conversations and their sections.

Why can I not find a Conversation?
Confirm that the correct workspace is selected. You can only see Conversations your Remesh account has permission to access.

Why am I seeing a rate-limit message or slow responses?
Large batches of requests may reach the workspace or user rate limit. Wait before retrying and send fewer requests at one time.