browser-max-automation

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install the Playwright MCP server using npx @playwright/mcp@latest, which downloads content from the official npm registry.\n- [COMMAND_EXECUTION]: Detailed instructions are provided for launching Microsoft Edge with the --remote-debugging-port=9222 flag to allow CDP-based automation.\n- [REMOTE_CODE_EXECUTION]: The tool browser_evaluate is documented for executing arbitrary JavaScript code within the browser context to manipulate the DOM.\n- [DATA_EXFILTRATION]: Use of the CDP endpoint enables the agent to interact with existing browser sessions, potentially accessing sensitive information like cookies and active login states.\n- [PROMPT_INJECTION]: The skill is inherently vulnerable to indirect prompt injection (Category 8) as it processes data from untrusted external websites.\n
  • Ingestion points: Web page content is ingested through browser_navigate and browser_snapshot (SKILL.md).\n
  • Boundary markers: The skill does not implement or recommend the use of delimiters to isolate untrusted web content from instructions.\n
  • Capability inventory: The skill allows for high-impact actions including simulated user input (browser_type), navigation, and arbitrary script execution (browser_evaluate) (SKILL.md).\n
  • Sanitization: No sanitization or validation logic for external web content is described in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 09:24 AM