executing-plans
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, unauthorized network access, or credential exposure were detected. The skill follows best practices by requiring critical review of input plans and establishing checkpoints for human verification before proceeding with task batches.- [SAFE]: The skill includes explicit instructions to avoid modifying sensitive branches (main/master) without consent and to stop immediately if blockers or ambiguities are encountered.- [SAFE]: Analysis of the indirect prompt injection surface: Ingestion point is Step 1 (SKILL.md) which reads external plans; Boundary markers are absent; Capability inventory includes implicit plan execution tasks (file/command execution); Sanitization is handled through mandatory critical review and human feedback checkpoints.
Audit Metadata