Remesh MCP: Setup Guide

 

Included in this Article

• Getting Started
• Anthropic: Claude Setup
• OpenAI: GPT Setup
• Microsoft: Co-Pilot Setup
• Verifying the Connection
• Frequently Asked Questions

Getting Started

To learn more about what the Remesh MCP server is, please reference this article.

A Remesh admin must navigate to the “Manage Workspace” > “Settings” area in the lower left corner of the Remesh dashboard.

1. Find the “MCP Access” tab
2. Enable the workspace toggle or both workspace and team toggles

Note, if GenerativeAI is disabled, MCP Access is also disabled by default.

Anthropic: Claude Setup

Custom connectors are available on all Claude plans (Free plans are limited to one custom connector), but who can add one depends on your account type:

• Personal accounts (Free / Pro / Max): you can add the connector yourself — follow Steps 1–4 below.
• Team / Enterprise accounts: the Add button and Add custom connector option are not shown to regular members. Only an organization Owner (or Primary Owner) can add custom connectors, and they do it at the organization level — see Team / Enterprise setup below. Once the Owner has added it, it appears in every member's Connectors list and each member just clicks Connect.

Don't see "Add" in your Connectors settings? You're on a Team/Enterprise account and aren't an Owner. Send this guide to your Claude workspace admin and skip to Step 4 once they've added the connector.

Team / Enterprise setup (Owner required)

On Team and Enterprise plans, custom connectors are managed at the organization level. Regular members won't see the Add button at all — an Owner or Primary Owner adds the connector once for the whole organization:

1. Go to Organization settings → Connectors.
2. Click Add, hover over Custom, and select Web.
3. Enter the remote MCP server URL: https://live.remesh.chat/mcp (leave the OAuth Client ID/Secret in Advanced settings blank).
4. Click Add.

After the Owner adds it, each member connects individually: go to Settings → Connectors, find the Remesh connector (labeled Custom), click Connect, and sign in with your Remesh credentials. Adding the connector makes it available to the organization, but no one gets access until they authenticate with their own Remesh account — access stays scoped to what each user can already see in Remesh.

Note for Claude admins: connectors added this way are visible to all members of the organization (per-user/per-group connector gating isn't currently available on Team plans). Enterprise plans with role-based permissions can manage this more granularly.

Personal account setup (Free / Pro / Max)

Step 1 — Open Connector settings

In Claude, click your profile icon → Settings, then select Connectors in the left sidebar (under Customize).

Claude Settings — Connectors page

Claude Settings — Connectors page

Step 2 — Add a custom connector

Click the Add button in the top right, then choose Add custom connector from the dropdown.

Add dropdown — Add custom connector

Add dropdown — Add custom connector

Step 3 — Enter the Remesh MCP details

In the Add custom connector modal:

Click Add.

Add custom connector modal — name and server URL

Add custom connector modal — name and server URL

Step 4 — Connect and sign in

Remesh now appears in your Connectors list. Click Connect and sign in with your Remesh credentials when prompted. Once authorized, Claude can use the Remesh tools in any chat — try asking: “List my Remesh workspaces.”

OpenAI: GPT Setup

ChatGPT requires a few more steps because unverified MCP apps must be added through Developer Mode and published at the workspace level before individual users can enable them. You’ll need workspace admin/owner permissions for Steps 1–3. Developer Mode is available on Pro, Plus, Business, Enterprise, and Edu accounts on the web.

The flow is: enable Developer Mode → create the app in Workspace Settings → publish the draft → add the published app to your own account.

Step 1 — Enable Developer Mode (admin)

1. Click your profile → Settings (your personal settings, not workspace settings).
2. Go to Security and login.
3. Toggle Developer mode on. You’ll see an “Elevated risk” label — this is ChatGPT’s standard warning for unverified connectors and is expected.

Profile menu — open personal Settings

Profile menu — open personal Settings

Security and login — Developer mode toggle

Security and login — Developer mode toggle

If you don’t see the Developer mode toggle, a workspace admin needs to grant developer permissions in your workspace’s admin settings first.

Step 2 — Create the app in Workspace Settings (admin)

1. Click your profile → Workspace settings (this opens chatgpt.com/admin).
2. Select Apps in the left sidebar.
3. Click + Create (top right). The New App modal opens.

Profile menu — open Workspace settings

Profile menu — open Workspace settings

Workspace settings — Apps page with Create button

Workspace settings — Apps page with Create button

1. Fill in the New App modal:

1. Check “I understand and want to continue” under the Custom MCP servers introduce risk warning (expected for any app not yet reviewed by OpenAI), then click Create.

New App modal — server URL, OAuth, and risk acknowledgment

New App modal — server URL, OAuth, and risk acknowledgment

Step 3 — Publish the draft (admin)

The new app is created as a draft — it isn’t available to anyone until it’s published.

1. Close the modal and return to the Apps page in Workspace settings.
2. Click the Drafts tab.
3. Find Remesh MCP and click Publish.

Apps — Drafts tab where the new app is published

Apps — Drafts tab where the new app is published

Step 4 — Add the published app to your account (each user)

Publishing makes the app available to the workspace, but each user still adds it to their own account:

1. Leave Workspace settings and go back to your personal Settings.

2. Find Remesh App in your workspace's available apps and enable/add it.

3. Sign in with your Remesh credentials when prompted (OAuth).

Once connected, Remesh MCP appears in your Plugins list with a green Connected status. Click Try in Chat or just start a new conversation and ask: "List my Remesh workspaces."

Remesh MCP connected — Plugins page with Try in chat

Microsoft: Co-Pilot Setup

Setup in Microsoft Copilot (Copilot Studio)

Copilot works differently from other agents like Claude and ChatGPT: there's no end-user "add a connector to chat" option. Instead, MCP servers are attached as tools on a Copilot Studio agent — you (or your Copilot Studio maker/admin) add the Remesh MCP to an agent, and that agent can then be used in Copilot Studio, Microsoft Teams, or Microsoft 365 Copilot once published. This is typically done by whoever builds agents for your team.

This flow follows Microsoft's official documentation (linked below and in Sources). Field names may vary slightly as Copilot Studio evolves — Microsoft's connect an existing MCP server page is the authoritative reference.

Before you start:

• You need access to Copilot Studio with permission to edit an agent (maker role in your Power Platform environment).
• The agent must have generative orchestration turned on (required for MCP).
• MCP connections run through Power Platform connectors, so your organization's data loss prevention (DLP) policies apply — if the connection is blocked, a Power Platform admin needs to allow it.

Step 1 — Open your agent's Tools page

In Copilot Studio, open (or create) the agent you want to connect to Remesh, then go to its Tools page.

Step 2 — Add the Remesh MCP server

• Select Add a tool → New tool → Model Context Protocol. The MCP onboarding wizard opens.
• Fill in the server details:

• Click Create, then Next.

The server description matters more here than in other clients — the agent's orchestrator uses it to decide when to call Remesh at runtime, so keep it clear and specific.

Step 3 — Create the connection and sign in

In the Add tool dialog, select Create a new connection, sign in with your Remesh credentials when prompted, then select Add to agent.

Step 4 — Review the tools

The Remesh MCP now appears on the agent's Tools tab. Open it to see the full list of Remesh tools it provides. All tools are enabled by default (Allow all); if you want the agent to use only a subset — for example, analysis tools but not conversation editing — turn off Allow all and toggle individual tools.

Test it in the agent's test pane: "List my Remesh workspaces."

Verifying the connection

In any client, try these prompts:

• "Ping the Remesh MCP to check connectivity."
• "List my Remesh workspaces."
• "What were the top themes by agreement in [conversation name]?"

Analysis answers include [N] citations that link back to the source view in your Remesh dashboard.

Frequently Asked Questions

Why do I see an "unverified" or "risk" warning during setup? The Remesh MCP isn't yet listed in the official Anthropic/OpenAI connector directories (verification is in progress), so both clients treat it as a custom connector and show their standard warning. The connection itself is authenticated with your Remesh login and scoped to your existing Remesh permissions.

I don't see the "Add" button or "Add custom connector" in Claude. This is expected on Team/Enterprise accounts: custom connectors are managed at the organization level, and only an Owner or Primary Owner sees the option (under Organization settings → Connectors). Regular members can't add connectors themselves — ask your Claude workspace admin to add the Remesh connector for the organization, then it will appear in your own Settings → Connectors list where you click Connect and sign in. On personal accounts (Free/Pro/Max), the Add button is always available.

I don't see the Developer mode toggle in ChatGPT. Developer mode requires admin permissions (and on Enterprise/Edu, an admin may need to grant developer access to your role). It's also only available on paid plans, on the web.

I published the app in ChatGPT but can't use it. Publishing makes it available to the workspace — you still need to add/enable the app in your personal Settings (Step 4) and sign in with your Remesh account.

I can't find where to add MCP in Microsoft Copilot. MCP servers aren't added in the Copilot chat app — they're added to an agent in Copilot Studio (Tools → Add a tool → New tool → Model Context Protocol). If you don't have Copilot Studio access, ask whoever manages agents in your organization.

The Copilot Studio connection is blocked or the MCP option is missing. Two common causes: the agent doesn't have generative orchestration enabled (required for MCP), or a Power Platform DLP policy is blocking the connector — MCP access in Copilot Studio runs through Power Platform connectors, so a Power Platform admin may need to allow it. See Microsoft's MCP troubleshooting page.

Sign-in fails or tools return "not authorized." Access mirrors what you can see in Remesh. Confirm you can log in at Remesh directly with the same account, and that the account has access to the workspace you're querying.

Requests are being throttled. The Remesh MCP is rate limited per workspace and per user. Wait a moment and retry.