Analystes en assurance qualité des logiciels et testeurs Deep adversarial verification of code changes, specifically optimized for catching AI-generated code issues that pass superficial tests but fail in production. Use this skill whenever the user wants to verify code, review AI-generated code, audit a PR or diff for hidden bugs, check edge cases, or validate code correctness beyond basic tests. Triggers on: "verify this code", "check this PR", "audit this diff", "review this change for bugs", "is this code production-ready", "deep review", "adversarial review", "find issues in this code", "check for AI bugs", "validate this implementation", "security audit this code", "edge case analysis". Also trigger when user is about to merge a PR, when they express uncertainty about code correctness, or when they say things like "looks right to me but can you double-check". The 66% of developers frustrated with "almost right" AI code need this — do not skip it when code quality verification is at stake.
2026-07-14