Get the FREE Ultimate OpenClaw Setup Guide →

Direwolf

Distributed Data Processing Pipeline for MCP.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio framebuffers-direwolf docker run -i framebuffers-direwolf \
  --env DW_CONFIG="path/to/config.json" \
  --env DW_NET_PORT="12345" \
  --env DW_LOG_LEVEL="info"

How to use

Direwolf is a data extraction, manipulation and transformation framework designed to work with BIM applications via an MCP-based server. It implements the Wolfpack data transport and provides components for querying, storing and visualising BIM model insights. The server is intended to run in a containerized environment (via Docker) and exposes an MCP-compatible interface so clients and AI agents can request model data, parameters, and derived insights. Use it to connect Revit or other BIM tools to the Wolfden RAM database and the Hunter MCP server, enabling fast, in-memory data operations and JSON-RCP 2.0 based communication.

How to install

Prerequisites:

  • Docker installed and running on your host
  • A suitable environment to host the Direwolf MCP server (recommended: Linux or Windows with Docker Desktop)

Installation steps:

  1. Ensure Docker is running and you have network access to pull images.
  2. Pull the Direwolf MCP server image (if a public image is published): docker pull framebuffers-direwolf
  3. Run the container to start the MCP server: docker run -d -p 8080:8080 -e DW_CONFIG=/config/sample.json -e DW_LOG_LEVEL=info --name direwolf framebuffers-direwolf
  4. (Optional) mount a local configuration file if the image supports it, e.g.: docker run -d -p 8080:8080 -v /path/to/config.json:/config/sample.json -e DW_CONFIG=/config/sample.json --name direwolf framebuffers-direwolf
  5. Verify the server is up by querying its MCP endpoint or checking container logs: docker logs -f direwolf

If you build from source or use another packaging format, follow the repository-specific build steps provided by the authors.

Additional notes

Notes:

  • This MCP server is described as experimental and in active development. Expect breaking changes and evolving features.
  • The repository describes components: Direwolf (host), Wolfden (RAM database), Hunter (MCP server) and Wolfpack (JSON-RCP 2.0 data format). The MCP driver for Direwolf may be in development, so some MCP capabilities might be limited in this release.
  • When running in Docker, you may need to expose additional ports or mount configuration files depending on your environment and workflow.
  • Include the DLLs or libraries required by your BIM application if you integrate with Revit, as suggested by the project notes. Ensure the container has access to any needed resources and that the URIs for internal resources are correctly configured.
  • If you encounter issues, check the DW_LOG_LEVEL environment variable and container logs for diagnostic messages.

Related MCP Servers

Sponsor this space

Reach thousands of developers