mcpserver.so LogoMCPFly
Submit
Weights & Biases MCP Server logo

Weights & Biases MCP Server

by wandb

AI/MLwandbweavetracingmonitoring

The Weights & Biases MCP Server is a Model Context Protocol server designed for querying Weights & Biases Weave traces. It enables MCP Clients to query W&B Weave traces and write text and charts to W&B Reports.

View on GitHub

Last updated: N/A

What is Weights & Biases MCP Server?

The W&B MCP Server is a server that allows querying of Weights & Biases Weave traces using the Model Context Protocol. It provides tools for querying traces, counting traces, and creating W&B reports.

How to use Weights & Biases MCP Server?

To use the server, first install it by cloning the repository, setting up a virtual environment with uv, and installing the package. Configure the server by creating a .env file with your WANDB_API_KEY. Then, run the server using uv run src/mcp_server/server.py. Finally, configure your MCP client (e.g., Claude Desktop) to connect to the server.

Key features of Weights & Biases MCP Server

  • Query W&B Weave traces

  • Write text and charts to W&B Reports

  • Execute GraphQL queries against W&B experiment tracking data

  • Efficiently count Weave traces with filtering

Use cases of Weights & Biases MCP Server

  • Analyzing model performance using W&B Weave traces

  • Creating shareable W&B Reports with analysis findings and generated charts

  • Integrating W&B data into LLM-powered applications via MCP

  • Querying experiment tracking data including Projects, Runs, Artifacts, Sweeps, Reports, etc.

FAQ from Weights & Biases MCP Server

What is the purpose of the MCP Server?

The MCP Server allows querying Weights & Biases Weave traces and writing text and charts to W&B Reports.

How do I install the MCP Server?

Clone the repository, set up a virtual environment with uv, and install the package using uv pip install -e ..

How do I configure the MCP Server?

Create a .env file in the root directory with your WANDB_API_KEY.

How do I run the MCP Server?

Run the server using uv run src/mcp_server/server.py.

What tools are available in the MCP Server?

The server provides tools for querying W&B data with GraphQL, querying Weave traces, counting Weave traces, and creating W&B reports.