ArangoDB MCP Server

by MCP-Mirror

280 downloads Not rated yet
GitHub

About

The ArangoDB MCP Server is an implementation of the Model Context Protocol (MCP) for ArangoDB, enabling AI models to query and interact with ArangoDB databases. It provides tools for executing AQL queries, listing databases and collections, and reading documents via resource…

Explore

- Execute read‑only AQL queries with readQuery
- Execute any AQL query (including writes) with readWriteQuery
- List all databases on an ArangoDB server with listDatabases
- List all collections in a database with listCollections
- Read documents by database, collection, and ID via resource templates
- Client pool – one client per database for efficient connections

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

  1. Download and install Highlight from highlightai.com/download
  2. Navigate to the plugins tab and select "Add Custom Plugin"
  3. Configure the plugin with the settings below
    Plugin Name ArangoDB MCP Server
    Command (node, npx, python, etc.)

    Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.

  4. Enable "Start Automatically" if you want the plugin to start when Highlight launches

From the repository

To connect to an arangodb instance running on localhost:2434, to the database "account", add the following to your claude_desktop_config.json, assuming the path to this project is /home/yourcoolname/arango-mcp-server:

{
  "mcpServers": {
    "arangodb-account": {
      "command": "npx",
      "args": [
        "-y",
        "arango-mcp-server",
        "http://localhost:8529",
        "root",
        "root"
      ]
    }
  }
}

readQuery

Execute read-only query on the database

databaseName

(string): The database to query

aql

(string): The read-only AQL query to execute

readWriteQuery

Execute query on the database

listDatabases

List all the databases on the ArangoDB server

listCollections

List all the collections in an ArangoDB database

- readQuery
- Execute read-only query on the database
- Input:
- databaseName (string): The database to query
- aql (string): The read-only AQL query to execute
- Returns: Query results as array of objects
- readWriteQuery
- Execute query on the database
- Input:
- databaseName (string): The database to query
- aql (string): The AQL query to execute
- Returns: Query results as array of objects
- listDatabases
- List all the databases on the ArangoDB server
- Returns: Array of the databases names
- listCollections
- List all the collections in an ArangoDB database
- Input:
- databaseName (string): The name of the database
- Returns: Array of objects { "name": "<collectionName>" }

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "arangodb mcp server": {
            "arangodb-account": {
                "command": "npx",
                "args": [
                    "-y",
                    "arango-mcp-server",
                    "http://localhost:8529",
                    "root",
                    "root"
                ]
            }
        }
    }
}

McpServers

{
    "arangodb-account": {
        "command": "npx",
        "args": [
            "-y",
            "arango-mcp-server",
            "http://localhost:8529",
            "root",
            "root"
        ]
    }
}

This is an implementation of the Model Context Protocol for ArangoDB.

Overview

To be filled.

Components

Resources

Tools

Query Tools

- readQuery
- Execute read-only query on the database
- Input:
- databaseName (string): The database to query
- aql (string): The read-only AQL query to execute
- Returns: Query results as array of objects
- readWriteQuery
- Execute query on the database
- Input:
- databaseName (string): The database to query
- aql (string): The AQL query to execute
- Returns: Query results as array of objects
- listDatabases
- List all the databases on the ArangoDB server
- Returns: Array of the databases names
- listCollections
- List all the collections in an ArangoDB database
- Input:
- databaseName (string): The name of the database
- Returns: Array of objects { "name": "<collectionName>" }

Usage

To connect to an arangodb instance running on localhost:2434, to the database "account", add the following to your claude_desktop_config.json, assuming the path to this project is /home/yourcoolname/arango-mcp-server:

{
  "mcpServers": {
    "arangodb-account": {
      "command": "npx",
      "args": [
        "-y",
        "arango-mcp-server",
        "http://localhost:8529",
        "root",
        "root"
      ]
    }
  }
}

Development

Clone the repository.
Install everything.
Setup the dev environment.
Run the watcher.
Edit index.ts.

$ npm install
$ npm run dev:setup
$ npm run dev

Go to http://localhost:5173/ to see the inspector.

Todo

- [ ] Properly study the spec to see if the current implementation of resources actually make sense (I don't think it does)
- [x] The resource templates make sense
- [ ] Change all the "arango" to "arangodb" (repo name included...)
- [ ] Add back the arangodb password
- [ ] Proper README
- [ ] Tools/resource/etc following the format of the official anthropic stuff
- [ ] Figure out notifications
- [ ] Health checks
- [ ] More tools?
- [ ] Access all the databases running on an arangodb instance
- [ ] Release on npm somehow so it can be used with npx
- [ ] resources/subscribe and notifications/resources/list_changed and resources/unsubscribe
- [x] Properly document tools in the readme
- [x] Like on the SQLite MCP client
- [x] write_query tool separated from read_query -> actually is readWriteQuery
- [x] list_collections (see list_tables)
- [x] Client pool ie one client per database
- [x] Dev environment
- [x] resources/read with a template to read any document by database name, collection, id.
- [x] Add username and passwords as parameters of the command

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.