browser-tools
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious behavior or patterns were detected. The skill focuses on enhancing the security and ethics of browser automation.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell and browser commands safely. It emphasizes the use of environment variables for credentials and 'chmod 600' for sensitive files to prevent unauthorized access.
- [EXTERNAL_DOWNLOADS]: The skill defines a process for checking 'robots.txt' via 'curl' on target domains. This is a standard procedure for maintaining ethical scraping standards and is considered a safe external reference.
- [DYNAMIC_EXECUTION]: The skill utilizes 'agent-browser eval' to run JavaScript in the browser for page analysis (e.g., checking for status codes). This execution is confined to the browser environment and follows established safety patterns.
- [PROMPT_INJECTION]: The skill operates on untrusted web content, creating an inherent surface for indirect prompt injection. 1. Ingestion points: agent-browser commands (get text, snapshot). 2. Boundary markers: Absent. 3. Capability inventory: Browser interaction, shell execution, and file system access. 4. Sanitization: Absent. No malicious instructions were detected in the skill's own logic.
Audit Metadata