Roam Research MCP Server

Roam Research MCP Server is an MCP server that connects AI assistants like Claude to your Roam Research graph via its backend REST API. Designed for researchers, writers, and knowledge workers, it enables LLMs to query notes, create pages, update blocks, and organize graph structures without requiring the Roam Desktop application to run locally. Because it communicates directly with Roam's cloud backend using a graph name and graph API token, the server operates headlessly in server environments, CI pipelines, Docker containers, and local development setups. In addition to serving model context over stdio or HTTP daemon modes, the package functions as a standalone command-line interface with standard input piping support for terminal automation. AI clients can perform full-text graph searches, traverse block references and breadcrumbs, discover namespace subpages, and safely update pages using markdown diffing that preserves existing block UIDs. Built-in privacy controls respect standard tags like rm-hide and rm-private to withhold sensitive data from AI interactions.

Category: Notes & Knowledge Management

Tags: graph database, knowledge-management, notes, roam research

Visit Roam Research MCP Server

How to install and configure Roam Research MCP Server

1. Install the package globally via npm: npm install -g roam-research-mcp 2. Configure your environment variables with your Roam graph credentials, such as your graph name, graph API token, and optional multi-graph settings like ROAM_GRAPHS. 3. Add the server configuration to your MCP client config (such as Claude Desktop or Cursor). Set the command to roam and pass server or the relevant stdio flags, or use npx -y roam-research-mcp. 4. Save your configuration and restart your MCP client to access Roam graph tools.

What you can do with Roam Research MCP Server

• Search graph contents across blocks, namespaces, or specific pages to retrieve context and linked references directly into an AI workspace. - Append thoughts, structured meeting notes, or piping terminal command outputs directly into daily pages or designated project documents. - Perform non-destructive page updates using markdown diff algorithms that preserve underlying block UIDs and reference integrity. - Query task statuses to locate incomplete TODO items or finished tasks across entire graph hierarchies. - Build nested block structures using multi-graph configurations and protected write keys in headless server or CI environments.

Key facts

• https://github.com/2b3pro/roam-research-mcp
• Notes & Knowledge Management
• graph database, knowledge-management, notes, roam research

Part of MCP Servers

How do I install Roam Research MCP Server?

You can install the tool globally through npm using the command npm install -g roam-research-mcp. Once installed, it provides both the roam standalone terminal command and the MCP server interface for AI agents.

What can Roam Research MCP Server do?

The server exposes tools for reading pages, searching graph text, querying block references, creating pages with mixed content, updating blocks, and inspecting subpage namespaces. It also includes diffing logic to preserve block UIDs during page edits.

How does this differ from the official Roam MCP server?

The official server communicates with Roam Desktop's local HTTP API and requires the desktop application to be actively running. This server communicates directly with Roam's backend REST API, allowing headless operation in containers, remote servers, and automation scripts without Desktop.

Is Roam Research MCP Server open source?

Yes, Roam Research MCP Server is open source and distributed under the MIT license on GitHub, allowing free usage, modification, and integration into custom workflows.

Does Roam Research MCP Server support deleting pages?

No, the server deliberately omits a page deletion tool because Roam lacks an undo mechanism for bulk API deletions. This conservative design prevents accidental permanent data loss from automated agent actions.