The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the lighthouse command-line tool via the Python subprocess module. It follows security best practices by passing arguments as a list rather than a single shell string, which prevents shell injection vulnerabilities even when processing user-provided URLs.
[EXTERNAL_DOWNLOADS]: The skill depends on the external lighthouse tool (provided by Google) to fetch and audit web pages. The network operations performed are limited to the intended purpose of auditing the target URLs specified by the user.
[SAFE]: A thorough review of the Python scripts and skill metadata confirms that the skill operates transparently as described. No malicious patterns, obfuscation, hardcoded credentials, or unauthorized data exfiltration attempts were detected.

lighthouse-audit