239.AIAI post-training datasets and custom data for model and agent teams. Technical insights on coding, terminal, multimodal, and synthetic data.https://239.ai/enCoding Agent Training Data Quality: Why Passing Tests Isn't Enoughhttps://239.ai/en/posts/coding-agent-training-data-quality/https://239.ai/en/posts/coding-agent-training-data-quality/Why can a coding patch pass tests and still be wrong? A closer look at task specifications, verifier errors, reward reliability, and the limits of executable feedback in SWE RL.Thu, 12 Nov 2026 00:00:00 GMTCoding AgentsTraining Data QualityEvaluation & Verification