expectations-checklist

Operator intent is reconciled by storing acceptance criteria as a structured expectations checklist for AI-assisted tasks. This prevents inconsistent results by preserving human-readable wishes tied to PRD requirements instead of relying on agent interpretations that may drift over time.

Evidence-typed QA verification enforces per-item evidence requirements categorized as empirical, static, or measurement. This ensures QA checks during /dr-qa are specific and testable before an acceptance criterion item can be marked as met.

When a checklist item is partial or missed, fail-block routing stops the pipeline until an operator provides a valid override. Without an override, routing is blocked and the task routes back to /dr-do with the focused wish-ids to address the specific failures.

Yes, you can amend an expectations checklist without losing history because the system uses append-only evolution. It supports canonical-first creation and later append-merge amendments, ensuring the checklist evolves without rewriting or losing prior operator intent and verification records.

Yes, the expectations checklist requires task-scoped YAML frontmatter for compliance gating. This structured metadata is essential for reconciling operator intent during /dr-compliance checks across PRD- and plan-based routing workflows.