diagnosing-bugs

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides purely instructional content for debugging methodologies. It encourages the use of standard development tools such as tests, curl, CLI invocations, and headless browsers.
  • [COMMAND_EXECUTION]: The skill mentions executing various commands (curl, tests, scripts) and includes a helper bash script scripts/hitl-loop.template.sh. These are standard development activities intended to be run by the user or agent in a local development environment to reproduce bugs and do not involve unauthorized privilege escalation or persistence.
  • [EXTERNAL_DOWNLOADS]: The skill references external tools like Playwright and Puppeteer as options for building feedback loops. These are well-known, trusted industry-standard tools for UI testing and do not represent a security risk.
  • [DATA_EXFILTRATION]: The hitl-loop.template.sh script captures user input and prints it to the console for the agent to parse. This is a legitimate functional requirement for human-in-the-loop interaction and does not involve exfiltrating sensitive data to unauthorized external domains.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 12:22 PM
Security Audit — agent-trust-hub — diagnosing-bugs