Skip to Content
Welcome to 10ex Docs — explore Tutorials, Guides, Reference, Concepts, Use Cases, and the Agent Store.
ReferenceREST APIagents
Reference

agents

Base path: /api/v1/agents

The agents module is how you work with the Agent Store programmatically. You browse the catalog, hire an agent into a workspace, kick off runs, fetch results, and observe runtime behavior (cost, latency, Langfuse traces). Resources you’ll touch include agent definitions, hires, runs, run outputs, traces, and approval-gated previews. Schemas live in app/modules/agents/schemas.py.

What you can do with the agents API

  • List agents in the store and filter by category or capability
  • Hire an agent into your workspace (binds it to your connectors and brand knowledge)
  • Trigger a run with the agent’s typed input schema
  • Poll a run for status, partial output, and final result
  • Fetch usage and cost telemetry for billing reconciliation
  • List currently active agents in your workspace
  • Pull Langfuse trace IDs to debug a specific execution

Common operations

  • List the public Agent Store catalog and inspect a single agent’s I/O schema
  • Hire an agent (a one-time setup that wires connectors and knowledge)
  • Trigger an execution with a typed payload, receiving a run ID back
  • Poll or stream the run until it completes
  • Retrieve the structured output (or the artifact IDs it produced in assets, videos, marketing)
  • Cancel an in-flight run
  • List recent runs filtered by agent, status, or time window

Authentication & scoping

Workspace-scoped via API key in the Authorization: Bearer tnx_... header. Every run is attributed to the calling workspace and the calling key. JWTs from the user-login flow work too.

Pagination & filtering

List endpoints paginate with cursor-based limit + next_cursor. See Pagination for the shared envelope.

How agents authentication works

Agents inherit the caller’s workspace scope. Hires, runs, and outputs only appear if your API key belongs to the workspace that owns them. See Authentication for the full credential model.

MCP equivalents

If you’re driving 10ex from an LLM, the same surface is available over MCP.

MCP prompt

list_agents_in_store()

MCP prompt

get_agent_details({ slug: "email-sequence-generation" })

MCP prompt

list_active_agents()

Last updated on
userscrm