Agents

Run an autonomous task with the Kriya agent. The agent plans, calls the tools you allow, and returns the full step trace plus a final result.

POST https://api.setu.dwet.ai/v1/agents/run

Request body

Field	Type	Description
`model`	string	Agent model ID, e.g. `kriya-v1`.
`task`	string	The objective for the agent to accomplish.
`tools`	string[]	Allowed tool ids: `web_search`, `browser`, `code_runner`, `file_io`, `http_request`.

Example request

bash

curl https://api.setu.dwet.ai/v1/agents/run \
  -H "Authorization: Bearer $SETU_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kriya-v1",
    "task": "Research the top 3 UPI providers and summarize their fees",
    "tools": ["web_search", "code_runner"]
  }'

Example response

json

{
  "id": "run-abc123",
  "object": "agent.run",
  "created": 1782033025,
  "model": "kriya-v1",
  "status": "completed",
  "steps": [
    { "index": 0, "kind": "plan", "title": "Plan", "detail": "..." },
    { "index": 1, "kind": "tool", "title": "Call Web Search", "tool": "web_search", "detail": "..." },
    { "index": 2, "kind": "observation", "title": "Observation", "detail": "..." },
    { "index": 3, "kind": "final", "title": "Final answer", "detail": "..." }
  ],
  "output": "Task complete...",
  "usage": { "prompt_tokens": 18, "completion_tokens": 120, "total_tokens": 138 }
}

Each step has a kind of plan, tool, observation or final. Runs are billed by tokens against the project that owns the key. See Models for agent model IDs.