Skip to content

No documentation for using Browser from Python without Robot Framework #5164

Description

@aaltat

Browser keywords can be called from a plain Python script with no Robot Framework run anywhere, and argument conversion makes this a supported way to use the library. There is no documentation for it.

Two behaviours are specific to this context, and both are already measured:

  • browser.outputdir is the current working directory. Browser/browser.py:829 falls back to self._output_dir when EXECUTION_CONTEXTS.current is empty, so screenshots, videos, traces and the Playwright log land wherever the script was started from.
  • validate and then raise RobotNotRunningError: Cannot access execution context. All 31 keywords taking an AssertionOperator still accept and convert the operator — only these two operators' execution needs a Robot Framework run. This is pre-existing and outside the library's control.

Neither holds under Robot Framework. In particular they do not hold when a user's own library owns the Browser instance: Robot Framework is still running, so ${OUTPUTDIR} resolves and BuiltIn().evaluate works. Conflating "no Robot Framework" with "not imported by Robot Framework" is the mistake this issue exists to prevent.

Where it belongs

One closing section on /docs/extending/browser-as-a-base, once that page exists. The two pages covering the Robot Framework-hosted cases are specced in the site repository as docs/superpowers/specs/2026-08-16-python-extension-docs.md, and this case is explicitly out of scope there.

Parking it costs nothing today: the argument-conversion behaviour is documented by those two pages regardless, because it applies to any call reaching a keyword through attributes.

Metadata

Metadata

Assignees

No one assigned

    Type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions