Native browser automation with typed Python APIs and before-and-after evidence.
Documentation 路 Get started 路 Examples 路 PyPI 路 Issues
Control browsers from Python with the native
agent-browser engine.
Inspect a page, act on an element, and get the before-and-after snapshots
with a diff of what changed.
Alpha: The public API and native integration can change between minor releases. Pin a version for production automation.
python -m pip install pyagentbrowserRequires Python 3.11 through 3.14. Local browsers run headlessly by default. The first launch finds Chrome or Chromium, or downloads Chrome for Testing when needed.
from agentbrowser import Browser
with Browser.launch() as browser:
browser.page.set_content(
"<button onclick=\"this.textContent='Done'\">Run</button>"
)
snapshot = browser.observe()
result = snapshot.one(role="button", name="Run").click()
print(result.after.one(role="button").name)
print(result.diff.changed)Done
True
observe() captures a Snapshot of the page's accessible elements. Selecting
a button returns a Ref. Clicking it returns an ActionResult with before,
after, and diff.
Use browser.find for live element queries.
- Inspect and act. Navigate pages, fill forms, read documents, and capture screenshots and PDFs through typed namespaces.
- Manage sessions. Launch a browser or attach to an existing one. Work with tabs, cookies, saved state, and explicit cleanup.
- Control actions. Configure domain restrictions and confirmation policy for operations through the native engine.
- Compose in Python. Use synchronous or asynchronous controllers, agent
skills, and the complete engine protocol through
browser.native.
Get started 路 Concepts 路 Action evidence 路 Safety 路 API reference 路 Troubleshooting
The documentation site covers browser workflows, optional packages, supported platforms, and integrations. See examples for Python scripts and Contributing for development setup and checks.
pyagentbrowser is licensed under the Apache License 2.0. See NOTICE for bundled software attribution.