Tasks
Structured AI functions you can run standalone or compose into agents.
A Task is a structured AI function: a prompt plus typed inputs and a typed output, running on a model you choose. Agents call Tasks as tools, but Tasks are first-class over the API on their own — many integrations only ever call Run Task.
These endpoints let you create and version tasks, run them with typed input, list and inspect runs, and wire up the surrounding machinery: Task Forwarders (deliver task output to webhooks or email), Input Processors (pre-process input before a run), and Task Evals (offline evaluation).
All endpoints are scoped to an organization and project and require a bearer token or OAuth client credentials. See Authentication.
Key endpoints
Related
- Concept guide: Tasks
- Calling tasks from your own code: Run tasks via API
- Delivering output: Webhooks & forwarders