delivery-preflight-gate
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is entirely instructional markdown and does not include any executable scripts, binaries, or configuration that would trigger unauthorized actions.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local repository commands for verification (e.g., build, test, lint). It includes specific constraints to avoid excessive resource usage and requires human approval for any actions touching secrets, authentication, billing, or destructive behavior.
- [PROMPT_INJECTION]: No prompt injection or behavior override patterns were detected. The skill defines clear boundaries and anti-patterns that reinforce agent safety and user oversight, specifically instructing the agent not to replace human approval with automated gates for sensitive operations.
Audit Metadata