Verification Before Completion

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is purely instructional, focusing on best practices for software development verification and honesty in reporting completion status.
  • [PROMPT_INJECTION]: While the skill uses strict language like 'The Iron Law' and 'Non-negotiable' to define its workflow, it does not attempt to bypass system safety filters, override core agent instructions, or extract system prompts.
  • [DATA_EXFILTRATION]: There are no commands or instructions that access sensitive files or perform network operations.
  • [COMMAND_EXECUTION]: The skill mentions common developer tools like git, linters, and test runners in an illustrative context for verification, but does not execute unauthorized or hidden scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 08:58 AM
Security Audit — agent-trust-hub — Verification Before Completion