Get started
Quick start
Run your first browser agent in a few minutes — a plain-language task, a managed browser, structured output, and a signed webhook when it finishes.
1. Install the SDK
Install the BrowserPilot SDK and set your API key, which you can mint in the dashboard under Settings → API Keys.
pip install browserpilot-sdk export BROWSERPILOT_API_KEY=your_key
2. Create a run
A run takes a task in plain language and an optional output shape. The agent plans the steps, drives a managed browser, and returns typed JSON.
import asyncio
from browserpilot import BrowserPilot
async def main():
client = BrowserPilot()
run = await client.runs.create(
task="Open the careers page and list every open role",
output={"roles": [{"title": "", "team": "", "location": ""}]},
)
print(run.output)
asyncio.run(main())3. Watch it work
Every run streams live in a real browser. Open the run in the dashboard to watch it step by step, or replay it later — each step keeps its screenshot, action, and reasoning.
4. Receive the webhook
Register an endpoint in Settings → Webhooks. BrowserPilot delivers a signed event for every run; verify the x-browserpilot-signature header with your endpoint secret, then handle the payload.
POST /your-endpoint
x-browserpilot-signature: t=1715731200,v1=...
{
"event": "run.completed",
"run_id": "run_9fA2c1",
"steps": 14,
"cost_usd": 0.27,
"output": { "roles": [ ... ] }
}Next steps
Read Core concepts to understand runs, sessions, and cost, then the REST API reference for the full surface.