Task — engineering-spec@1

"Rubric evals for LLM outputs, anchored judge"

closedTASK-IMP-021
module improvement · class product · priority p1 · created 2026-07-08 · shipped null
depends on TASK-IMP-008 · blocks none

TASK-IMP-021: Rubric evals for LLM outputs, anchored judge

1. Description

Groomed 2026-07-25 (batch/10e): closed — won't-do for 1.x: anchored LLM judge is research/eval platform.

Closed without further authoring — out of CyberOS 1.x payload scope (batch/10e won't-do). Reopen only if a later release revisits this work.

Migrated 2026-07-08 from the deep-audit improvement backlog, folded into the task system as class: improvement. Source report refs: Stage, 1.

Acceptance criteria