MongoDB Atlas Managed MCP Server

Name: MongoDB Atlas Managed MCP Server
Author: MongoDB

MongoDB's hosted MCP endpoint for querying Atlas clusters and administering Atlas projects

Official · MongoDB TypeScript Remote (hosted) · Streamable HTTP Local · stdio Databases Cloud platforms

Website

44 popularity score ★ 1.2k repo ↓ 177/wk

Connect your client

Remote (hosted)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server.

Or add it by command or config

Claude CodeTerminal

claude mcp add --transport http mongodb-atlas-managed https://mcp.mongodb.com/

Claude (desktop and web)Settings → Connectors → Add custom connector

https://mcp.mongodb.com/

ChatGPTSettings → Apps → Advanced → Developer mode, then Create app

https://mcp.mongodb.com/

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "mongodb-atlas-managed": {
      "url": "https://mcp.mongodb.com/"
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "mongodb-atlas-managed": {
      "type": "http",
      "url": "https://mcp.mongodb.com/"
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.mongodb-atlas-managed]
url = "https://mcp.mongodb.com/"

Gemini CLITerminal

gemini mcp add --scope user --transport http mongodb-atlas-managed https://mcp.mongodb.com/

Local (npm)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

It needs MDB_MCP_API_CLIENT_ID, MDB_MCP_API_CLIENT_SECRET. Replace each <NAME> with your value.

Claude CodeTerminal

claude mcp add mongodb-atlas-managed -e MDB_MCP_API_CLIENT_ID=<MDB_MCP_API_CLIENT_ID> -e MDB_MCP_API_CLIENT_SECRET=<MDB_MCP_API_CLIENT_SECRET> -- npx -y mongodb-atlas-mcp-remote@latest

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "mongodb-atlas-managed": {
      "command": "npx",
      "args": [
        "-y",
        "mongodb-atlas-mcp-remote@latest"
      ],
      "env": {
        "MDB_MCP_API_CLIENT_ID": "<MDB_MCP_API_CLIENT_ID>",
        "MDB_MCP_API_CLIENT_SECRET": "<MDB_MCP_API_CLIENT_SECRET>"
      }
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "mongodb-atlas-managed": {
      "command": "npx",
      "args": [
        "-y",
        "mongodb-atlas-mcp-remote@latest"
      ],
      "env": {
        "MDB_MCP_API_CLIENT_ID": "<MDB_MCP_API_CLIENT_ID>",
        "MDB_MCP_API_CLIENT_SECRET": "<MDB_MCP_API_CLIENT_SECRET>"
      }
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "mongodb-atlas-managed": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "mongodb-atlas-mcp-remote@latest"
      ],
      "env": {
        "MDB_MCP_API_CLIENT_ID": "<MDB_MCP_API_CLIENT_ID>",
        "MDB_MCP_API_CLIENT_SECRET": "<MDB_MCP_API_CLIENT_SECRET>"
      }
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.mongodb-atlas-managed]
command = "npx"
args = ["-y", "mongodb-atlas-mcp-remote@latest"]

[mcp_servers.mongodb-atlas-managed.env]
MDB_MCP_API_CLIENT_ID = "<MDB_MCP_API_CLIENT_ID>"
MDB_MCP_API_CLIENT_SECRET = "<MDB_MCP_API_CLIENT_SECRET>"

Gemini CLITerminal

gemini mcp add --scope user -e MDB_MCP_API_CLIENT_ID=<MDB_MCP_API_CLIENT_ID> -e MDB_MCP_API_CLIENT_SECRET=<MDB_MCP_API_CLIENT_SECRET> mongodb-atlas-managed npx -- -y mongodb-atlas-mcp-remote@latest

Settings 4

`MDB_MCP_API_CLIENT_ID`	Required: Client ID of the service account Atlas creates for an MCP configuration (npm bridge only).
`MDB_MCP_API_CLIENT_SECRET`	Required: Secret for that client ID, created with the MCP Configuration API (npm bridge only).
`MDB_MCP_API_BASE_URL`	Optional: Endpoint the bridge forwards to; https://mcp.mongodb.com by default.
`MDB_MCP_TOKEN_TIMEOUT_MS`	Optional: Timeout for the token request in milliseconds; 10000 by default.

What it does

The Atlas Managed MCP Server is the version of the MongoDB MCP Server that MongoDB runs for you at https://mcp.mongodb.com. Nothing runs on your machine: an AI client connects to the endpoint and works against clusters hosted in MongoDB Atlas, both their data (queries, aggregations, inserts, updates, indexes, collections) and the Atlas account around them (organizations, projects, clusters, database users, alerts, Performance Advisor, Stream Processing).

There are two ways in. With user-delegated access, a person signs in with OAuth from a client MongoDB has registered (Claude Code, Claude Desktop and web, ChatGPT, Codex, Cursor, Devin and a few others) and the client acts with that person's own Atlas roles. With programmatic access, an administrator creates an MCP configuration through the Atlas Admin API; Atlas issues a service account for it, and an agent connects through the small npm bridge `mongodb-atlas-mcp-remote`, which reads `MDB_MCP_API_CLIENT_ID` and `MDB_MCP_API_CLIENT_SECRET`, exchanges them for a token at cloud.mongodb.com and forwards stdio traffic to the hosted endpoint.

The tool set is not the same as the self-run mongodb-mcp-server. MongoDB's tools table marks 41 tools as available on the managed server: it adds `remote-atlas-connect`, `aggregate-db` and `atlas-load-sample-dataset`, and leaves out `connect`, `atlas-connect-cluster`, `atlas-create-access-list`, `update-one`, `export`, `mongodb-logs`, `switch-connection` and the `atlas-local-*` tools. In read mode, write tools are not registered at all, so the client does not see them.

What it needs

To run it yourself	Node.js

A MongoDB Atlas account with clusters hosted in Atlas. For OAuth sign-in: a client from MongoDB's supported list, an Organization Owner who has turned on AI client access under App Connections (off by default except for organizations created at sign-up), and one of the Atlas roles allowed to authorize clients. For the npm bridge: Node.js with npx, plus a client ID and secret from an MCP configuration created through the Atlas Admin API (the docs use Atlas CLI 1.58.0 or later).

Limits

Works only with Atlas-hosted clusters; Community or Enterprise deployments need the self-run server. Rate limits are 2,000 requests per minute per IP and 500 per user; MongoDB says they may change without notice and cannot be raised on request. OAuth access lapses after 7 days idle or 30 days from grant, then the user signs in again. AI client access is set per organization: no per-user, per-cluster or per-client controls, and owners cannot revoke one user's connection. Tools that write files or need a Voyage AI key locally (`export`, embedding generation) are not offered.

Who it suits

Suits Atlas users who want an assistant to explore and change their cluster data and Atlas resources without running anything locally, and teams that want actions tied to an Atlas user or a dedicated service account in the audit log. Pick the self-run mongodb-mcp-server (which we also list) for MongoDB outside Atlas, for local Atlas deployments, for tools such as `export` or `mongodb-logs`, for a client MongoDB has not registered without setting up an MCP configuration, or when traffic would exceed the hosted rate limits.

Tools 41

atlas-list-orgs: List the Atlas organizations you can see
atlas-list-projects: List Atlas projects
atlas-create-project: Create an Atlas project
atlas-list-clusters: List Atlas clusters
atlas-inspect-cluster: Show details of one cluster
atlas-create-free-cluster: Create a free-tier cluster
atlas-get-regions: List Atlas regions for AWS, Azure or GCP
atlas-upgrade-cluster: Move a cluster from Free or Flex to a larger tier
atlas-create-cluster: Create a dedicated M10 to M80 cluster
atlas-pause-resume-cluster: Pause or resume a dedicated cluster
atlas-inspect-access-list: Show the IP and CIDR ranges allowed to reach a cluster
atlas-list-db-users: List database users
atlas-create-db-user: Create a database user
atlas-list-alerts: List a project's alerts
atlas-get-performance-advisor: Fetch Performance Advisor suggestions for a cluster
atlas-load-sample-dataset: Load MongoDB's sample data into a cluster
remote-atlas-connect: Open a connection to an Atlas cluster and return a handle for later calls
atlas-streams-discover: Inspect Stream Processing workspaces, processors and connections
atlas-streams-build: Create Stream Processing workspaces, processors or connections
atlas-streams-manage: Start, stop or update Stream Processing resources
atlas-streams-teardown: Delete Stream Processing resources
find: Query documents in a collection
aggregate: Run an aggregation pipeline on a collection
aggregate-db: Run an aggregation pipeline at database level
count: Count documents in a collection
explain: Show the query plan the optimizer picked
insert-many: Insert documents into a collection
create-index: Create an index, including vector search indexes
drop-index: Remove an index
update-many: Update every matching document
rename-collection: Rename a collection
create-collection: Create a collection
delete-many: Delete matching documents
drop-collection: Delete a collection
drop-database: Delete a database
list-databases: List databases on the connection
list-collections: List collections in a database
collection-indexes: Show a collection's indexes
collection-schema: Infer a collection's schema
collection-storage-size: Report a collection's size in MB
db-stats: Report database statistics

Popularity

44 popularity score, on our list (the bar is 36)

★ 1.2k repo ↓ 177/wk

GitHub stars, whole repository	1,158	18 points
npm downloads, last week	177	18 points
Official server	Yes	8 points

It lives in the folder packages/mongodb-atlas-mcp-remote of mongodb-js/mongodb-mcp-server, which holds more than this server. Its stars are the whole repository's, so in the score they never count for more than its own downloads.

Measured 2026-11-07. How the score works.

Where it is published

Code	mongodb-js/mongodb-mcp-server/packages/mongodb-atlas-mcp-remote ★ 1.2k, the whole repository
npm	mongodb-atlas-mcp-remote 1.0.0, 2026-08-10
Hosted	`https://mcp.mongodb.com/` Streamable HTTP · OAuth sign-in

Also in mongodb-js/mongodb-mcp-server: MongoDB MCP Server, mongodb-atlas-mcp-remote.

What it can reach

Runs	On your machine, or on the maker's servers
Reaches	A database, Cloud resources, Your account at a service, The internet
Changes things	Yes: it can write, send, create or delete
Sign-in	OAuth sign-in for the hosted endpoint; an API key or token to run it yourself
Licence	Apache-2.0 (npm)
Last code change	2026-08-10 (a release)

Code runs on MongoDB's servers; the npm bridge only relays stdio to https://mcp.mongodb.com. Tools can create clusters (paid M10+ tiers among them), database users, projects and Stream Processing resources, and can insert, update, delete or drop data. OAuth clients act with the signing-in user's Atlas roles, capped by the organization's read or read-write mode; MCP configurations act as their own service account, created read-only by default, with an optional IP access list. Atlas does not log individual tool calls; audit events they cause record the user and client or the service account. Turning access off ends control-plane access, but database users a client created keep working.

Over time

npm downloads a week 177 +4 (+2 %)

22812

20 Jul 20267 Nov 2026

Before the site measured it, read from npm's daily download counts.

Popularity score 44 +4 (+10 %)

4640

20 Jul 20267 Nov 2026

Before the site measured it, read from those stars and downloads, by today's rule.

What is rising across the directory

Sources

Page checked 2026-10-14.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![MongoDB Atlas Managed MCP Server on EveryMCPServer](https://everymcpserver.com/badge/mongodb-atlas-managed-mcp-server.svg)](https://everymcpserver.com/servers/mongodb-atlas-managed-mcp-server)

Connect your client

Add in one click

Or add it by command or config

Add in one click

Or add it by command or config

Settings 4

What it does

What it needs

Limits

Who it suits

Tools 41

Similar servers

MongoDB MCP Server

mongodb-atlas-mcp-remote

MCP DigitalOcean Integration

Azure MCP Server

Heroku Platform MCP Server

Confluent Cloud managed MCP server