wandb.ai
Weights & Biases is a machine learning development platform for experiment tracking, model evaluation, artifacts, observability, and related workflows. It also offers hosted inference, serverless training, and agent-facing tooling such as an MCP server.
wandb.ai exposes a GraphQL Public API, two authenticated HTTP APIs for Serverless Inference and Serverless Training, a hosted MCP server, and the `wandb` CLI; all documented developer surfaces use a W&B API key.
- W&B MCP Serverdiscovered
- W&B Serverless Inference APIdiscovered
- W&B Serverless Training APIdiscovered
- W&B Public APIdiscovered
- wandb CLIdiscovered
Create a personal or service-account API key in W&B. The quickest path is Authorize or User Settings: sign in, create a new API key, and copy it immediately because W&B only shows the full key once. For service-account-owned keys, use the Service Accounts area described in the W&B Quickstart. Store it as WANDB_API_KEY or pass it as a bearer token where supported.
conventions · 0/8 published
- integrations.json✗
/.well-known/integrations.json - llms.txt✗
/llms.txt - API catalog✗
/.well-known/api-catalog - OpenAPI document✗
/api/schema/, /openapi.json, /swagger.json, /api/openapi.json, or /v1/openapi.json - MCP server card✗
/.well-known/mcp/server-card.json - OAuth protected resource✗
/.well-known/oauth-protected-resource - Agent card✗
/.well-known/agent-card.json - Agent skills✗
/.well-known/agent-skills/index.json
Publish these signals → /publishing