diagnosing-bugs
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill outlines a comprehensive debugging discipline focused on building feedback loops and verifying fixes. The instructions are purely pedagogical and do not contain malicious instructions or patterns.
- [COMMAND_EXECUTION]: The skill encourages the use of various development tools like git bisect, curl, and browser automation scripts (e.g., Playwright). These commands are standard for the intended use case of bug diagnosis and are executed within the agent's environment as part of its normal operation.
- [SAFE]: The provided hitl-loop.template.sh script is a simple bash template for interactive sessions with a user. It uses standard read commands to capture feedback and does not perform any sensitive operations or network requests autonomously.
- [SAFE]: The skill references reading local project documentation such as CONTEXT.md to establish a mental model of the codebase, which is a secure and standard practice for context gathering.
Audit Metadata