autoresearch-item-level-failure-vs-bad-prompt

A 100% failure rate on a specific test item is a broken test item — not evidence of a bad prompt. Discarding the prompt in response to item-level failures is a false signal. autoresearch v2.5.0 adds explicit item-level failure detection at experiment loop step 9 to flag these broken items before any discard decision. Distinguish item failure from prompt failure before triggering plateau logic.

Related

• item-level-failure-detection-separates-prompt-from-test-item
• 2026-04-04-oracle-001-self-architecture-analysis
• autoresearch-v2-5-0-upgrade-8-gaps-absorbed
• clawteam-openclaw-multi-agent-swarm-evaluation
• autoresearch-validation-set-prevents-score-noise