Agentic coding without the cloud: evaluating open-weight large language models on longitudinal data preparation tasks

[system]

---

This is a companion discussion topic for the original entry at https://arxiv.org/abs/2607.21482