The Enterpret MCP server brings your customer feedback — themes, accounts, sentiment, verbatim quotes — into Claude, Cursor, Notion, and Glean. This article covers what it can do, how to set it up, and how to troubleshoot a connection.

Key capabilities:

Query your organization's complete customer feedback dataset
Get cited answers where every aggregate number links back to source records in the Enterpret dashboard
Pull verbatim customer quotes with speaker attribution
Discover entities, taxonomy values, and source-specific metadata fields on demand

Setup Instructions

Claude (Web & Desktop)

Note: Connectors require a paid Claude plan (Pro, Max, Team, or Enterprise).

Pro and Max

Click here to add Enterpret to Claude, then click Connect to authorize access.

...or, find it manually:

Open Claude and click Customize in the sidebar
Click Connectors
Click the + button
Search "Enterpret"
Click Connect to authorize access

Ask your question directly — no initialization step is required.

Team and Enterprise

Setting up Enterpret for your team is a two-step process: an admin adds the connector for the organization, then each team member connects individually.

Step 1: Admin adds the connector

Click here to add Enterpret to Claude, then click Add to enable it for your organization.

...or, find it manually:

Navigate to Admin settings > Connectors
Search for "Enterpret" in the Connectors Directory and click Add

Then:

Navigate to Settings > Connectors
Find Enterpret in the list and click "Connect" to authorize access

Once connected, the Enterpret MCP becomes available to every member in your organization — no need to set it up user by user.

Step 2: Members connect individually

Click here to add Enterpret to Claude, then click Connect to authorize access.

...or, find it manually:

Navigate to Settings > Connectors
Find Enterpret in the list of available connectors
Click "Connect" to authorize access

Ask your question directly — no initialization step is required.

Each member authenticates with their own account — queries and usage are tracked individually.

Claude Code

Open Terminal (or Command Prompt on Windows) and run:

claude mcp add --transport http --scope user enterpret https://wisdom-api.enterpret.com/server/mcp

Start Claude Code:
```
claude
```
Run the /mcp command, select Enterpret from the list of MCP servers, then click "Authenticate" to complete the authentication flow

Ask your question directly — no initialization step is required.

Cursor

Quick Install (Recommended)

In your Enterpret dashboard, navigate to Settings > Enterpret MCP
Click the Connect button next to Cursor
This will start the configuration process in Cursor with your organization's MCP settings prefilled

Ask your question directly — no initialization step is required.

Manual Setup

If auto-configuration doesn't work, you can manually configure Cursor:

Press Cmd+Shift+P (Mac) or Ctrl+Shift+P (Windows) to open the command palette
Type "Cursor Settings" and select it
Navigate to MCP in the left sidebar
Click Add Custom MCP

Add the following configuration:

{
  "mcpServers": {
    "enterpret": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://wisdom-api.enterpret.com/server/mcp"]
    }
  }
}

Save the file (Cmd+S or Ctrl+S)

Ask your question directly — no initialization step is required.

Codex

Connect Enterpret to Codex to query customer feedback directly from your Codex workspace. Codex connects over Streamable HTTP with a Bearer token.

In Codex, open Settings and select MCP server
Click Add and enter:
- Name: Enterpret
- Mode: Streamable HTTP
- URL: https://wisdom-api.enterpret.com/server/mcp
Add your Bearer token in the token field. To generate one, expand the Bearer Token Authentication section below (Settings > Enterpret MCP > Generate under Auth Token > Copy).
Save. Ask your question directly — no initialization step is required.

Notion Custom Agents

Build Notion Custom Agents that run on top of your customer feedback — scheduled digests, in-page @mentions on PRDs, and database-backed workflows. Notion supports the full MCP protocol, so Enterpret works out of the box.

Follow the step-by-step setup in Enterpret in Notion Custom Agents.

Note: Requires a Notion Business or Enterprise plan. A workspace admin must enable Custom MCP servers under Settings → Notion AI → AI connectors (one-time).

Glean

Connect Enterpret to Glean so users can query customer feedback from the Glean Assistant and wire Enterpret tools into Glean Agent Builder workflows. Glean is admin-managed — a Glean workspace admin sets up the connection once in the Admin Console, and every authorized user gets access to it.

In Glean, go to Admin Console → Platform → Actions, click Add, and choose Import tools from MCP server. Use the following configuration:

MCP server URL: https://wisdom-api.enterpret.com/server/mcp
Transport type: Streaming HTTP
Authentication: OAuth (recommended) or Bearer token from Settings → MCP in your Enterpret dashboard

Once connected, an admin can enable individual Enterpret tools for the Glean Assistant and for Agent Builder, and apply role-based access to control which creators can use them in agents. For the full admin walkthrough, see Glean's documentation: Connect Remote MCP Servers to Glean.

Other AI Tools

For MCP-compatible AI tools not listed above, use this URL to connect to the Enterpret MCP: https://wisdom-api.enterpret.com/server/mcp

Ask your question directly — no initialization step is required.

Bearer Token Authentication

In addition to OAuth, Enterpret MCP supports Bearer token authentication for MCP clients that require this method.

Navigate to Settings > Enterpret MCP in your Enterpret dashboard
Click Generate under Auth Token
Copy the token
Add the token to your MCP configuration using the --header flag:

--header "Authorization: Bearer YOUR_AUTH_TOKEN_HERE"

Example configuration:

{
  "mcpServers": {
    "enterpret": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://wisdom-api.enterpret.com/server/mcp",
        "--header",
        "Authorization: Bearer YOUR_AUTH_TOKEN_HERE"
      ]
    }
  }
}

Note: Tokens expire after 6 months. Generate a new token and update your configuration when needed.

Available Tools

Your AI assistant picks the right tools automatically based on what you ask — you don't need to call them by name. They're listed here so the names look familiar when you see them appear in your chat interface.

get_organization_details — Identifies your Enterpret organization, so answers link back to your dashboard.
get_graph_schema — Lets the assistant see what kinds of feedback you track and how it's organized.
get_query_examples — Gives the assistant reference patterns for common kinds of feedback questions before it answers.
search_graph_fields — Surfaces integration-specific data points like Zendesk tags, App Store ratings, or Gong call metadata.
search_graph_values — Looks up the exact themes, accounts, sources, or taxonomy values in your data, so the assistant searches and filters on real labels.
run_graph_query — Pulls the feedback that answers your question, with citation links back to source records.
find_user_quote — Surfaces verbatim customer quotes from the records behind your answer.

Migrating from older tool names? The legacy names (get_schema, execute_cypher_query, search_knowledge_graph) keep working for the lifetime of your current session, so an in-flight conversation never breaks mid-stream. The new names take over the next time your host refreshes its tool list — typically when you start a new conversation or reconnect the MCP server. No manual migration is required.

Citations and quotes

Enterpret MCP grounds its answers in your underlying data. When the assistant returns a number or a claim, the evidence travels with it.

Citations on every claim

Every aggregate number in an answer — a count, a percentage, a top-N list — is a link. Click it to open the underlying records in your Enterpret dashboard, filtered to exactly what drove that number. It's the fastest way to verify a finding and to share evidence with a teammate.

Customer quotes in their own words

When you ask "what are customers saying about X?", your assistant returns actual verbatim text — the words customers used, with speaker attribution, plus a link back to the source feedback. No paraphrasing, no AI-written summary. If no validated quote exists for your request, your assistant will tell you so rather than making one up.

FAQs

I don't see the Enterpret MCP in my AI tool

Ensure you completely restarted your application (not just minimized)
Check that Node.js is installed by running node --version in Terminal
Verify you're using the correct MCP URL: https://wisdom-api.enterpret.com/server/mcp
Make sure you have a stable internet connection

I'm getting "Cannot connect to MCP server" error

Double-check your MCP URL - ensure you're using https://wisdom-api.enterpret.com/server/mcp
Ensure there are no extra spaces in your configuration
Try restarting your AI tool again
Verify you can access https://wisdom-api.enterpret.com/server/mcp in your browser

I'm getting "npx command not found" error (Windows)

Restart your computer after installing Node.js
Try using the full path: "C:\\Program Files\\nodejs\\npx.cmd" instead of "npx"

I'm behind a corporate firewall - what do I need?

Ensure your firewall allows access to npm registry (for downloading mcp-remote)
HTTPS traffic to mcp.enterpret.com must be allowed
If behind a corporate proxy, npm and HTTPS requests need proxy configuration

My older tool calls are returning a "DEPRECATED:" prefix message

The tool you're using has been renamed. The legacy name keeps working in your current session, but the new name will be what you see the next time your host refreshes its tool list. See the Available Tools list above for the mapping.

My MCP connection looks connected but isn't showing the new tools

MCP hosts cache the tool list when you first connect. To see the new tools, start a fresh conversation (or, in Claude Code or Cursor, disconnect and reconnect the Enterpret MCP server). Your current session keeps working with the legacy tool names in the meantime — there's nothing to fix.

I have access to multiple organizations - how does this work?

Each Enterpret organization uses its own API token to connect to multiple orgs, add a separate mcp-remote entry per org (same URL, unique name, port, and token) in claude_desktop_config.json. See Connecting Claude to Multiple Enterpret Organizations for the full template and steps.

Introduction to Enterpret

Enterpret Features Explained

Enterpret MCP with Claude Plugins

Connecting Claude to Multiple Enterpret Organizations

Enterpret in Notion Custom Agents