Browser Automation as a Service
Type a task in natural language and the AI agent will execute it in a real browser. Click any example below to try it.
Use your API key (found in Settings) to call the API programmatically.
# Create a task
curl -X POST https://your-host/api/task \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{"prompt": "Search for ..."}'
# Check task status
curl https://your-host/api/task/TASK_ID \
-H "Authorization: Bearer YOUR_API_KEY"
# Continue a completed task
curl -X POST https://your-host/api/task/TASK_ID/continue \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{"message": "Now also check ..."}'
# Cancel a running task
curl -X POST https://your-host/api/task/TASK_ID/cancel \
-H "Authorization: Bearer YOUR_API_KEY"
# List tasks (paginated)
curl "https://your-host/api/task?skip=0&limit=10" \
-H "Authorization: Bearer YOUR_API_KEY"
# Delete a task
curl -X DELETE https://your-host/api/task/TASK_ID \
-H "Authorization: Bearer YOUR_API_KEY"
Connect to ws://your-host/ws/task/TASK_ID to stream logs and status updates.
# Messages from server:
{"type": "log", "message": "...", "status": "running"}
{"type": "status", "status": "completed", "logs": [...]}
{"type": "ask_user", "question": "...", "mode": "text"}
{"type": "heartbeat"}
# Send to server:
{"type": "hitl_response", "response": "user answer"}
{"type": "cancel"}