browser-harness

Pass

Audited by Gen Agent Trust Hub on Oct 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local binaries such as uv, browser-harness, and chrome using Python's subprocess.run and subprocess.Popen. These calls are implemented using structured argument lists rather than shell strings, which effectively mitigates command injection risks.
  • [EXTERNAL_DOWNLOADS]: The skill automates the installation and upgrade of the browser-harness package using the well-known uv package manager. This is a standard and expected procedure for initializing the skill's core functionality.
  • [DATA_EXPOSURE]: The skill includes functionality for capturing screenshots and task recordings that might contain sensitive information. It implements a mandatory consent workflow to ensure the user is informed and explicitly opts in before recording features are enabled.
  • [DYNAMIC_EXECUTION]: The skill provides a run mechanism to execute Python code for browser control. The documentation includes clear instructions for the Agent to only execute trusted, self-generated code and to never execute untrusted scripts sourced from web pages.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 16, 2026, 03:39 AM