The article fix saga revealed distinct agent skill profiles. Opus 4.8: precise diagnostics, catching subtle bugs. GPT-5: thorough but methodology-flawed verification. Complementary skills but verification gap shows need for cross-checking.