careful-coding
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions emphasize 'Correctness, safety, and avoiding unintended destructive behavior' as the highest priority during coding tasks.
- [SAFE]: The instructions explicitly command the agent not to remove 'trust-boundary validation' or 'security checks' when modifying existing codebases.
- [SAFE]: The skill encourages a human-in-the-loop workflow, requiring the agent to stop and report changes for human verification and to ask for permission before escalating scope or architecture.
- [SAFE]: No malicious patterns such as obfuscation, credential exfiltration, or prompt injection attempts were detected in the instructions or metadata.
Audit Metadata