playwright-cli

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The run-code and eval commands allow for the execution of arbitrary JavaScript within the browser context, which is a powerful capability that could be abused if the agent is misled into running malicious scripts.\n- [COMMAND_EXECUTION]: The skill documentation includes a PowerShell command intended to persistently modify the host's PATH environment variable to facilitate tool installation, representing a persistent change to the user environment.\n- [PROMPT_INJECTION]: The skill facilitates the ingestion of data from arbitrary websites, creating a surface for indirect prompt injection.\n
  • Ingestion points: Page content retrieved through open, snapshot, and eval commands (SKILL.md).\n
  • Boundary markers: No delimiters or instructions are defined to separate untrusted web content from agent instructions.\n
  • Capability inventory: The tool can execute arbitrary scripts, interact with page elements, and manage browser sessions.\n
  • Sanitization: Content extracted from the browser is not sanitized before being processed by the agent.\n- [DATA_EXFILTRATION]: Documentation in references/storage-state.md details how to export and save sensitive browser storage state, including authentication cookies and tokens, to local files, which could lead to unauthorized access if these files are exposed.\n- [EXTERNAL_DOWNLOADS]: The skill relies on the installation of the playwright-cli package from the npm registry.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 11:17 PM