swe-debug-mantra
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's instructions strictly focus on defining a four-step debugging discipline (Reproduce, Know the fail path, Falsify the hypothesis, and Every run is a breadcrumb).
- [SAFE]: No network operations, sensitive file access, or credential harvesting patterns were detected.
- [SAFE]: The instructions for the agent to 'recite the mantra verbatim' and follow specific operating rules are functional constraints for the debugging process and do not constitute malicious prompt injection or behavior override.
- [SAFE]: No obfuscation techniques, such as Base64 encoding or hidden characters, were found in the text.
- [SAFE]: While the skill involves processing potentially untrusted data like stack traces or error logs (Category 8 surface), this is consistent with its primary function as a debugging tool and does not present an inherent security risk in this context.
Audit Metadata