ArcadeDB MCP Server
Expose ArcadeDB’s multi-model database (graph, document, vector, time-series) to your agent via the built-in stdio MCP server in the official Docker image.
Overview
com.arcadedb/mcp-server is a MCP server for the Build phase that provides built-in stdio access to ArcadeDB’s graph, document, vector, and time-series database.
What is this MCP server?
- Built-in MCP server shipping with ArcadeDB—no separate adapter repo required
- Multi-model support: graph, document, vector, and time-series in one engine
- stdio transport via OCI package docker.io/arcadedata/arcadedb:26.4.1-SNAPSHOT
- Open-source project hosted at github.com/ArcadeData/arcadedb
- Registry version 26.4.1-SNAPSHOT
- OCI package docker.io/arcadedata/arcadedb:26.4.1-SNAPSHOT with stdio transport
- Four modeled paradigms: graph, document, vector, time-series
Community signal: 931 GitHub stars.
What problem does it solve?
Agents cannot safely explore or script against your multi-model database without a first-class protocol bridge tailored to that engine.
Who is it for?
Builders prototyping graph+RAG or time-series features who standardize on ArcadeDB in Docker and want native MCP instead of ad-hoc SQL clients.
Skip if: Teams committed only to vanilla Postgres with no multi-model needs, or anyone unwilling to operate their own database container.
What do I get? / Deliverables
After you run ArcadeDB with MCP enabled over stdio, your agent can interact with the same database instance you use for backend development.
- Running ArcadeDB instance with MCP stdio attached to your agent
- Queryable multi-model data layer for backend and agent tooling
- Version-aligned deploy from registry package 26.4.1-SNAPSHOT
Recommended MCP Servers
Journey fit
Database design, queries, and agent-driven data modeling happen while you build the product backend, before production tuning dominates your week. Backend is the canonical shelf for primary data stores and query surfaces; ArcadeDB is the persistence layer agents can operate through MCP.
How it compares
Database-native stdio MCP server, not a generic SQL assistant skill or hosted serverless datastore.
Common Questions / FAQ
Who is com.arcadedb/mcp-server for?
Developers building on ArcadeDB who want Claude Code, Cursor, or Codex to run and reason about graph, document, vector, and time-series data through MCP.
When should I use com.arcadedb/mcp-server?
Use it during backend build-out when ArcadeDB is your primary store and you need agent-assisted modeling, queries, or integration tests against live data.
How do I add com.arcadedb/mcp-server to my agent?
Pull docker.io/arcadedata/arcadedb:26.4.1-SNAPSHOT (or matching release), configure your MCP client for stdio against the ArcadeDB process, and verify tools list in the agent.