alanscalone/llm-behavior-analysis

A 4-month multi-model evaluation analyzing LLM behavior, stress-testing constraints, and mapping sandbox dynamics.

3 best points · 0 comments · 1 Hacker News post · 2 GitHub stars · View on GitHub

Hacker News discussions

• A 400-hour forensic audit of LLMs using multi-model context saturation (3 points, 0 comments, by AlanScalone, 2026-05-26)