RushDB MCP Server

Name: RushDB MCP Server
Author: RushDB

Structured, graph-linked memory for AI agents, stored in a RushDB project

Official · RushDB TypeScript Remote (hosted) · Streamable HTTP Local · stdio Databases Knowledge and memory

GitHub npm Website

61 popularity score ★ 327 ↓ 702/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 rushdb https://mcp.rushdb.com/mcp

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

https://mcp.rushdb.com/mcp

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

https://mcp.rushdb.com/mcp

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "rushdb": {
      "url": "https://mcp.rushdb.com/mcp"
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "rushdb": {
      "type": "http",
      "url": "https://mcp.rushdb.com/mcp"
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.rushdb]
url = "https://mcp.rushdb.com/mcp"

Gemini CLITerminal

gemini mcp add --scope user --transport http rushdb https://mcp.rushdb.com/mcp

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 RUSHDB_API_KEY. Replace each <NAME> with your value.

Claude CodeTerminal

claude mcp add rushdb -e RUSHDB_API_KEY=<RUSHDB_API_KEY> -- npx -y @rushdb/mcp-server

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "rushdb": {
      "command": "npx",
      "args": [
        "-y",
        "@rushdb/mcp-server"
      ],
      "env": {
        "RUSHDB_API_KEY": "<RUSHDB_API_KEY>"
      }
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "rushdb": {
      "command": "npx",
      "args": [
        "-y",
        "@rushdb/mcp-server"
      ],
      "env": {
        "RUSHDB_API_KEY": "<RUSHDB_API_KEY>"
      }
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "rushdb": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@rushdb/mcp-server"
      ],
      "env": {
        "RUSHDB_API_KEY": "<RUSHDB_API_KEY>"
      }
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.rushdb]
command = "npx"
args = ["-y", "@rushdb/mcp-server"]

[mcp_servers.rushdb.env]
RUSHDB_API_KEY = "<RUSHDB_API_KEY>"

Gemini CLITerminal

gemini mcp add --scope user -e RUSHDB_API_KEY=<RUSHDB_API_KEY> rushdb npx -- -y @rushdb/mcp-server

Settings 4

`RUSHDB_API_KEY`	Required: RushDB API key from app.rushdb.com; a read-only key hides the write tools.
`RUSHDB_API_URL`	Optional: RushDB API base URL; defaults to https://api.rushdb.com/api/v1, set it for a self-hosted instance.
`MCP_TRANSPORT`	Optional: stdio by default; set to http to serve over HTTP with OAuth instead.
`PORT`	Optional: Port for HTTP mode; 3001 by default.

What it does

RushDB MCP Server connects an AI client to a RushDB project. RushDB is a graph database from the same makers: records carry labels and properties and can be linked to each other. Through the server, an assistant can save facts as records, change or delete them, link and unlink them, and query them later by property filters, by relationship traversal or by vector similarity. The makers present it as a persistent memory layer for agents that needs no hand-built schema or embedding pipeline.

There are two ways to use it. Locally, the npm package runs over stdio and talks to the RushDB API with an API key; RUSHDB_API_URL points it at a self-hosted RushDB or another instance instead of the default https://api.rushdb.com/api/v1. Hosted, the endpoint https://mcp.rushdb.com/mcp is meant for web clients such as ChatGPT and Claude.ai, and you sign in to your RushDB account through OAuth.

The README's setup text says there are 13 tools and its tables list 20, with PascalCase names. The source at version 2.13.0 (packages/mcp-server/tools.ts) registers 39 tools in camelCase. The extra tools cover schema discovery (getSchemaMarkdown, getSchema), embedding index management (createEmbeddingIndex, upsertEmbeddingVectors and others), vectorSearch, natural-language smartSearch, and property tools such as deleteProperty. semanticSearch is marked in the code as a deprecated alias of vectorSearch. At startup, the server looks up the API key's scopes and hides write tools when the key is read-only (index.ts). Setting MCP_TRANSPORT=http makes the same package serve over HTTP on PORT (3001 by default).

What it needs

To run it yourself	Node.js

Node.js with npx for the local package. You also need a RushDB account and an API key from app.rushdb.com, or a self-hosted RushDB instance set with RUSHDB_API_URL. For the hosted endpoint, you need a RushDB account to sign in with and a client that supports remote MCP with OAuth.

Limits

It works only on data in RushDB; it cannot reach other databases. The README's own counts (13 tools in the setup text, 20 in its tables, with PascalCase names) are out of date: version 2.13.0 registers 39 tools in camelCase (tools.ts). semanticSearch is a deprecated alias for vectorSearch. A read-only API key hides the write tools. The README warns that running npx @rushdb/mcp-server from a working directory inside the RushDB monorepo fails with MCP error -32000. The fix it gives is to pin a version: npx -y -p @rushdb/mcp-server@latest rushdb-mcp.

Who it suits

It suits people who already use RushDB, or who want an assistant to keep structured notes that link to each other across sessions and are willing to store them in RushDB's cloud or run their own RushDB instance. The publisher of the database maintains it. Web-client users can use the hosted endpoint with OAuth and install nothing. Pick something else if you want memory that stays on your own machine with no external service, or if your data lives in another database. Since it can delete records, properties and indexes, give it a read-only API key if the assistant only needs to read.

Tools 39

getSchemaMarkdown: Return the project's labels, counts and properties as compact Markdown
getSchema: Return the same schema as structured JSON
findLabels: List record labels and how many records each has
findProperties: List properties found across labels
findPropertyById: Get one property's name, type, label and record count
propertyValues: Get distinct values or statistics for one property
deleteProperty: Remove a property and its values from every record
createRecord: Store one new record
bulkCreateRecords: Store many records in one call
getRecord: Fetch a record by its ID
getRecordsByIds: Fetch several records by their IDs
findRecords: Query records by labels, filters, relationships or ordering
findOneRecord: Return the first record that matches a query
findUniqRecord: Return the one record matching a query, failing if there are zero or several
updateRecord: Change some fields of a record, keeping the rest
setRecord: Replace all fields of a record, dropping fields not given
deleteRecord: Delete records
deleteRecordById: Delete a single record by ID
bulkDeleteRecords: Delete every record matching a query
exportRecords: Export matching records as CSV
attachRelation: Link two records with a relationship
detachRelation: Remove a relationship between records
findRelationships: Search relationships by type, properties and endpoints
listRelationshipPatterns: Show inferred relationship patterns and analysis status
analyzeRelationshipPatterns: Queue an analysis that suggests relationship patterns
approveRelationshipPattern: Accept a suggested pattern and apply it
ignoreRelationshipPattern: Dismiss a suggested pattern
deleteRelationshipPattern: Delete a saved pattern, optionally with the links it created
findEmbeddingIndexes: List the project's embedding index policies and their status
createEmbeddingIndex: Create an embedding index on a text property
upsertEmbeddingVectors: Write your own vectors into an external embedding index
deleteEmbeddingIndex: Delete an embedding index and its stored vectors
getEmbeddingIndexStats: Show how many records an index has embedded so far
vectorSearch: Find records by vector similarity on an indexed property
semanticSearch: Deprecated alias of vectorSearch
smartSearch: Turn a plain-language question into a query, run it and return results
getSearchQuerySpec: Return the reference for RushDB's query syntax
getQueryBuilderPrompt: Return the built-in query-building instructions for clients without prompt support
helpAddToClient: Give setup steps for adding the server to an MCP client

Popularity

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

★ 327 ↓ 702/wk

GitHub stars	327	30.2 points
npm downloads, last week	702	22.8 points
Official server	Yes	8 points

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

Where it is published

Code	rush-db/rushdb ★ 327
npm	@rushdb/mcp-server 3.3.0, 2026-11-07
Hosted	`https://mcp.rushdb.com/mcp` Streamable HTTP · OAuth sign-in

What it can reach

Runs	On your machine, or on the maker's servers
Reaches	A database, 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	2026-11-07

Every tool calls the RushDB API: the cloud service at api.rushdb.com by default, or the instance set in RUSHDB_API_URL. Locally, the server runs on your machine with a RushDB API key. The hosted endpoint mcp.rushdb.com runs on RushDB's servers and uses OAuth with scoped tokens (records:read, records:write). Many tools write data, and some delete it permanently: deleteRecord, deleteRecordById, bulkDeleteRecords (deletes everything matching a query), deleteProperty (removes a field from every record), deleteEmbeddingIndex and setRecord (drops fields that are not given). At startup it reads the key's scopes and hides write tools from read-only keys. It does not touch local files or run shell commands.

Over time

GitHub stars 327 +2 (+1 %)

327325

13 Oct 20267 Nov 2026

npm downloads a week 702 +693 (+7700 %)

1.0k0

20 Oct 20257 Nov 2026

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

Popularity score 61 +16 (+36 %)

6140

13 Oct 20267 Nov 2026

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

[![RushDB MCP Server on EveryMCPServer](https://everymcpserver.com/badge/rushdb.svg)](https://everymcpserver.com/servers/rushdb)

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 39

Similar servers

MCPQL

ClickHouse MCP Server

ArcadeDB MCP Server

Memanto MCP Server

Metabase MCP server

Instant MCP