Runtime API — getting started
Call your deployed Agentman agent from your own code: base URL, agent-token authentication, and a first streaming request in under a minute.
Agents Developer Help
Every agent you build is also an API.
Call your deployed agents from your own backend, app, or website — stream responses in native LangChain format, upload files for your agent to read, and inspect its model and capabilities.
Developer API
Call your agent — the streaming API
POST /v3/agent/stream: the request body, every SSE event type, and working curl and JavaScript clients for streaming agent responses in native LangChain format.
Agent info, models, and capabilities
GET /v3/agent/info, /v3/agent/models, and /v3/agent/capabilities — everything your client needs to initialize: agent configuration, available models, and what file types the agent can read.
The Files API — upload, attach, manage
Upload files your agent can read, attach them to streaming calls, mint presigned URLs, and validate accessibility — the full /v3/files reference with examples.
Use your skills in Claude Managed Agents
Give Anthropic's Managed Agents access to your Agentman skills library — the two-part MCP config, the credential vault that signs it in, and the one piece Studio doesn't issue yet.