mcp-wandb
This MCP server lets an agent query Weights & Biases experiment data over stdio.
List W&B projects for a given entity.
List runs in a project, including run names, IDs, and states.
List unique metric names logged across a project's runs.
Plot selected metrics for a run and return a base64-encoded PNG.
Retrieve detailed run information: overview, config, summary, and system/environment metadata.
Allows querying Weights & Biases experiments, including listing projects, runs, metrics, retrieving run details, and plotting metrics as images.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-wandb list runs in my wandb project 'mcp-demo'"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Deprecated
This project is no longer maintained. Agents can now interact with Weights & Biases directly, so this dedicated MCP server is no longer needed.
mcp-wandb is a small Model Context Protocol server for querying Weights & Biases from MCP-compatible clients. It exposes W&B projects, runs, metrics, run details, and metric plots as FastMCP tools over stdio.
Use it when you want an agent to inspect experiment data without switching to the W&B dashboard or hand-copying run metadata.
Install
git clone https://github.com/tsilva/mcp-wandb.git
cd mcp-wandb
uv sync --locked --exclude-newer '7 days'
export WANDB_API_KEY=your_api_key
uv run python server.pyConfigure your MCP client to run the repo's server.py file, then restart the client.
{
"mcpServers": {
"wandb": {
"command": "python",
"args": ["/path/to/mcp-wandb/server.py"],
"env": {
"WANDB_API_KEY": "your_api_key"
}
}
}
}Related MCP server: metaflow-mcp-server
Commands
uv sync --locked --exclude-newer '7 days' # install the reviewed lockfile
uv run python server.py # run the MCP server over stdio
uv run pytest -q # run offline mocked behavior testsTools
get_wandb_projects(entity)lists projects for a W&B entity.list_wandb_runs(entity, project_name)lists run names, IDs, and states.list_project_metrics(entity, project_name)returns metric names found across runs.plot_run_metric(entity, project_name, run_id, metric_names)returns a PNG metric plot as a FastMCP image.get_run_details(entity, project_name, run_id)returns overview, config, summary, and system metadata.
Notes
Python 3.13 or newer is required.
WANDB_API_KEYmust be set before invoking a W&B tool. MCP tool discovery and startup do not contact W&B.The server uses
wandb.Apidirectly and does not keep a local database.Tests use mocked W&B responses, send no external requests, and exercise registration against the real MCP SDK.
The MCP SDK is held on its patched 1.x line because MCP 2 removes the FastMCP module used by this server.
Architecture

License
CI uses repository-selected, SHA-pinned setup actions. When rotating a pin, replace the corresponding allowlist entry with the reviewed release commit while retaining the selected-only policy and SHA pinning.
Available Tools
5 toolsget_run_detailsB
Retrieve detailed information about a W&B run, including:
Overview (name, id, state, etc.)
Config (user-defined hyperparameters)
Summary (final logged metrics)
System/environment details
Inputs: entity (required), run_id (required), project_name (required). Returns: result
get_wandb_projectsB
Get a list of projects from Weights & Biases.
Inputs: entity (required). Returns: result
list_project_metricsA
List all unique metric names logged in a W&B project.
Inputs: entity (required), project_name (required). Returns: result
list_wandb_runsB
List all runs for a given W&B project.
Inputs: entity (required), project_name (required). Returns: result
plot_run_metricB
Plot the requested metrics and return the base-64-encoded PNG.
Inputs: entity (required), run_id (required), metric_names (required), project_name (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool targets a distinct resource-action pair: projects, runs, run details, metric names, and plot output. There is no meaningful overlap or risk of confusion.
All names are lowercase snake_case and verb-first, but conventions are mixed: 'get' and 'list' are used inconsistently for collection-returning tools, and the 'wandb' prefix appears on only some tools. Still readable and predictable enough for an agent.
Five tools form a compact, well-scoped set for reading and visualizing W&B data. Each tool serves a clear purpose without redundancy.
Covers project discovery, run listing, run details, metric discovery, and plotting. However, raw metric history is only accessible through an image plot, and there is no run comparison or filtering, which are common W&B workflows.
Maintenance
Related MCP Connectors
Experiment tracking for coding agents: read your team's ML runs, metrics and lineage. Read-only.
Query your AI agent traces, root-cause verdicts and issues, and run investigations, from your agent.
Access Pollinations models and API capabilities through agent tools.
Web data for agents. Extract web pages, crawl websites, discover URLs, and search the web.
Related MCP Servers
- AlicenseCqualityDmaintenanceEnables AI assistants to interact with MLflow experiments, runs, and registered models. Supports browsing experiments, retrieving run details with metrics and parameters, and querying the model registry through natural language.7MIT
- AlicenseAqualityDmaintenanceEnables coding agents to interact with Metaflow workflows, including querying flows, runs, tasks, logs, and artifacts across any Metaflow backend.24Apache 2.0
- AlicenseNot gradedqualityDmaintenanceEnables querying W\&B Weave traces and creating W\&B Reports with text and visualizations through natural language.Apache 2.0
- AlicenseNot gradedqualityDmaintenanceQuery and analyze your Weights & Biases data using natural language through the Model Context Protocol.1MIT