debugging
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured workflow for debugging failures using causal evidence and systematic probing. It explicitly instructs the agent to redact secrets from shared outputs and captures, demonstrating a strong security-first approach.
- [COMMAND_EXECUTION]: The instructions involve executing symptom-specific tests and CLI harnesses to reproduce bugs. This is consistent with the primary purpose of a debugging skill and includes guidance to minimize the scope of execution and maintain acceptance targets.
- [INDIRECT_PROMPT_INJECTION]: The skill manages the risk of processing untrusted data (such as logs, HAR traces, and heap snapshots) by instructing the agent to focus on specific call paths and events rather than broad execution of external payloads. It mandates sanitizing sensitive fields and IDs before processing evidence.
- [DATA_EXFILTRATION]: No unauthorized network operations or exfiltration patterns were detected. The skill specifically warns against keeping sensitive payloads in logs and requires masking credentials.
Audit Metadata