Connect
Optimize
Secure
Live 27 Actions
Weights & Biases MCP Server
for AI Agents
Connect your AI agent to StackOne's Weights & Biases MCP server and give it 27 MCP tools out of the box. Auth, tool execution, and security all managed.
Weights & Biases MCP Server
Built by 
StackOne
StackOne 
Coverage
27 Agent Actions
Create, read, update, and delete across Weights & Biases — and extend your agent's capabilities with custom actions.
Authentication
Agent Tool Authentication
Per-user OAuth in one call. Your Weights & Biases MCP server gets session-scoped tokens with zero credentials stored on your infra.
Agent Auth →
Security
Agent Protection
Every Weights & Biases tool response scanned for prompt injection in milliseconds — 88.7% accuracy, all running on CPU.
Prompt Injection Defense →
Performance
Max Agent Context. Min Cost.
Free up to 96% of your agent's context window to enhance reasoning and reduce cost, on every Weights & Biases call.
Tools Discovery →What is the Weights & Biases MCP Server?
A Weights & Biases MCP server lets AI agents read and write Weights & Biases data through the Model Context Protocol — Anthropic's open standard for connecting LLMs to external tools. StackOne's Weights & Biases MCP server ships with 27 pre-built actions, fully extensible via the Connector Builder — plus managed authentication, prompt injection defense, observability, and agent execution runtime. Connect it from MCP clients like Claude Desktop, Claude Code, Cursor, Goose, and VS Code, or from agent frameworks like OpenAI Agents SDK, LangChain, and Vercel AI SDK.
All Weights & Biases MCP Tools
Every action from Weights & Biases's API, ready for your agent. Create, read, update, and delete — scoped to exactly what you need.
Artifact Collections
- Get Artifact Collection
Get a W&B artifact collection by name.
- List Artifact Collections
List artifact collections under a project's artifact type.
Artifact Types
- Get Artifact Type
Get a W&B artifact type by name within a project.
- List Artifact Types
List all artifact types within a project.
Jobs
- Get Job
Get a W&B Launch job by full artifact name.
- List Jobs
List W&B Launch jobs within a project.
Projects
- Create Project
Create a new W&B project within an entity.
- Get Project
Get a W&B project by name within an entity.
- List Projects
List W&B projects for an entity, with cursor pagination.
Runs
- Create Run
Create a new W&B run (shell) via the upsertBucket mutation.
- Get Run
Get a single W&B run by entity, project, and run id.
- List Runs
List W&B runs in a project with MongoDB-style filters and cursor pagination.
Teams
- Create Team
Create a new W&B team (entity).
- Get Team
Get a W&B team (entity) by name.
Users
- Create User
Provision a new W&B user account by email.
- Get User
Search for a W&B user by username or email (returns the first match).
- List Users
List W&B users, optionally filtered by username/email substring.
Other (10)
- Create Artifact (Empty)
Create a new empty W&B artifact (metadata-only, no files) via the GraphQL API.
- Create Custom Chart
Create a custom Vega chart preset for a W&B entity.
- Get Artifact
Get a W&B artifact by full name within a project.
- List Artifact Versions
List all versions of an artifact collection within a project.
- List Registries
List W&B Model Registries within an organization (Enterprise / Teams plans).
- List Reports
List W&B reports within a project.
- Get Sweep
Get a W&B hyperparameter sweep by id within a project.
- Get Viewer
Get the authenticated user (entity owner of the API key).
- Check Artifact Collection Exists
Check whether a W&B artifact collection exists.
- Check Artifact Exists
Check whether a W&B artifact exists.
Set Up Your Weights & Biases MCP Server in Minutes
One endpoint. Any framework. Your agent is talking to Weights & Biases in under 10 lines of code.
Agent Frameworks
JSON{
"mcpServers": {
"stackone": {
"command": "npx",
"args": [
"-y",
"mcp-remote@latest",
"https://api.stackone.com/mcp?x-account-id=<account_id>",
"--header",
"Authorization: Basic <YOUR_BASE64_TOKEN>"
]
}
}
}Platform Resources
MCP Code Mode: Keeping Tool Responses Out of Agent Context
Anthropic's code_execution processes data already in context. Custom MCP code mode keeps raw tool responses in a sandbox. 14K tokens vs 500.
11 min
Comparing BM25, TF-IDF, and Hybrid Search for MCP Tool Discovery
Benchmarking BM25, TF-IDF, and hybrid search for MCP tool discovery across 916 tools. The 80/20 TF-IDF/BM25 hybrid hits 21% Top-1 accuracy in under 1ms.
10 min
Indirect Prompt Injection Defense for MCP Tools: A Technical Guide
MCP tools that read emails, CRM records, and tickets are indirect prompt injection vectors. Here's how we built a two-tier defense that scans tool results in ~11ms.
12 min
Weights & Biases MCP Server FAQ
Does StackOne have a Weights & Biases MCP server?
Yes. StackOne offers a hosted Weights & Biases MCP server with 27 pre-built actions, and every action is tested and QA'd by StackOne. Connect it to Claude, Cursor, and any other MCP client, or to any agent framework through the AI Action SDK. It ships with managed agent authentication, prompt injection defense, and tool discovery with server-side execution that preserve your agent's context window and keep reasoning performance.
Weights & Biases MCP server vs direct API integration — what's the difference?
A Weights & Biases MCP server and direct API integration serve different use cases. Direct API integration is for software-to-software — backend code calling Weights & Biases. A Weights & Biases MCP server is for AI agents — MCP clients like Claude and Cursor, plus framework agents built with OpenAI, LangChain, or Vercel AI — discovering and calling Weights & Biases at runtime. StackOne provides both.
How does Weights & Biases authentication work for AI agents?
Weights & Biases authentication for AI agents works through a StackOne Connect Session. Create one via the dashboard or the SDK — you get an auth link and ready-to-paste config for Claude Desktop, Cursor, and other MCP clients. Your user authenticates their own Weights & Biases account; StackOne handles token exchange, storage, and refresh. Credentials never reach the LLM, and each user is isolated via
origin_owner_id.Are Weights & Biases MCP tools vulnerable to prompt injection?
Yes — Weights & Biases MCP tools can be vulnerable to indirect prompt injection. Any tool that reads user-written content — documents, messages, tickets, records, or free-text fields — is a potential vector. StackOne Defender scans every tool response before it enters the agent's context — regex patterns in ~1ms, then a MiniLM classifier in ~4ms. 88.7% accuracy, CPU-only.
What is the context bloat of a Weights & Biases agent and how do I avoid it?
Context bloat happens when Weights & Biases tool schemas and API responses eat your Weights & Biases agent's memory, preventing it from reasoning effectively. A single Weights & Biases query can return a massive JSON response, and connecting multiple tools compounds the problem. Tools Discovery and Code Mode reduce context bloat — loading only relevant tools per query and keeping raw responses out of the agent's context.
Can I limit which actions my Weights & Biases agent can access?
Yes — you can limit which actions your Weights & Biases agent can access directly from the StackOne dashboard. Toggle actions on or off, or restrict them to specific accounts, with no code changes to your agent. Session tokens can be scoped to exact actions so if one leaks, exposure stays contained.
Can I create custom agent actions for my Weights & Biases MCP server?
Yes — you can create custom agent actions for your Weights & Biases MCP server using Connector Builder. It's an integration agent your coding assistant (Claude Code, Cursor, or Copilot) can invoke to research Weights & Biases's API, generate production-ready connector YAML, test against the live API, and validate before you ship.
When should I NOT use a Weights & Biases MCP server?
Skip a Weights & Biases MCP server if your integration is purely software-to-software — direct Weights & Biases API integration is simpler when no AI agent is involved. For deterministic, compliance-critical operations (financial transactions, regulatory reporting), direct API gives you predictable behavior without agent-driven decision-making. MCP shines when AI agents need to dynamically discover and call Weights & Biases actions at runtime.
What AI frameworks and AI clients does the StackOne Weights & Biases MCP server support?
The StackOne Weights & Biases MCP server supports both. MCP clients (paste-and-go apps): Claude Desktop, Claude Code, Cursor, VS Code, Goose. Agent frameworks (code SDKs you build with): OpenAI Agents SDK, Anthropic, Vercel AI, Google ADK, CrewAI, Pydantic AI, LangChain, LangGraph, Azure AI Foundry.
Put your AI agents to work
All the tools you need to build and scale AI agent integrations, with best-in-class connectivity, execution, and security.