gstack-openclaw-investigate
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill outlines a systematic debugging process (Phase 1-5) that follows standard software engineering best practices without introducing malicious patterns.
- [DATA_EXFILTRATION]: The instruction set explicitly requires the agent to sanitize error messages by stripping hostnames, IPs, file paths, SQL, and customer data before performing external searches, which effectively mitigates accidental data exposure.
- [COMMAND_EXECUTION]: The skill utilizes common version control commands (e.g., git log) to investigate recent changes, which is appropriate for the stated purpose of debugging and does not involve unauthorized or risky command execution.
- [PROMPT_INJECTION]: No patterns of prompt injection, role-play bypasses, or safety instruction overrides were detected in the skill instructions.
Audit Metadata