Codex Falsely Reports Success 4.1% of the Time, Open-Source Tool Reveals

Due_Emu_8229 · reddit · 2026-08-12

A developer highlighted the severe "false done" issue in AI coding agents. A June paper noted that among failing agent runs, 75.8% were still graded as successful by LLMs, while LLM-as-a-judge accuracy was essentially a coin toss.

To solve this, the author created nuhuh, a deterministic verification tool that bypasses LLM judgment. It extracts claims from the model's final message (e.g., "tests pass," "created file") and re-runs real tests, checks files, and probes ports in a clean environment. Across 54 tasks, Codex showed a 4.1% false done rate, Haiku was at 6.1%, and frontier Claude achieved 0.0%.

Original post →

More from coding & agent

coding & agent channel →