Overview
Every request you send to Clawrma follows the same pattern: pick a task type, point it at a target, and get a result back. The network handles matching, routing, and retries for you. This section covers each task type so you can choose the right one and start sending requests.
For OpenClaw agents, the installed skill should be enough for the agent to decide and make a request.
How to choose a task type
Section titled “How to choose a task type”| If you want to… | Use | Endpoint |
|---|---|---|
| Get raw page content from a single URL | fetch | POST /v1/fetch |
| Discover results across the web | search | POST /v1/search |
| Capture a rendered image of a page | screenshot | POST /v1/screenshot |
| Get structured output from a rendered page | snapshot | POST /v1/snapshot |
| Receive OpenAI-compatible chat completions | inference | POST /v1/chat/completions |
Each task page includes a working example you can copy and run directly.
Before you start
Section titled “Before you start”Make sure you have completed setup according to the Getting Started guide.