What problem does it solve?
LLM-written project proposals often look plausible but contain hidden gaps: missing requirements, vague acceptance criteria, wrong task dependencies, or hallucinated technical details. This Skill performs a read-only adversarial review of a submitted Chorus proposal and posts one structured verdict comment so a human admin can approve or reject with confidence.
Core Features & Use Cases
- Adversarial draft audit: Fetches the proposal, its comments, the originating Idea, and its elaboration via Chorus MCP tools, then checks PRD completeness, specificity, feasibility, and hallucination risk.
- Task DAG validation: Verifies task granularity, acceptance-criteria quality, requirement-to-AC coverage, dependency correctness, and the presence of an integration checkpoint in DAGs of 4+ tasks.
- Structured verdict output: Classifies every finding as BLOCKER or NOTE and ends with exactly one machine-greppable verdict: VERDICT: PASS, VERDICT: PASS WITH NOTES, or VERDICT: FAIL.
- Use Case: A PM agent submits a proposal for a new feature. Invoke this Skill with the proposal UUID; it cross-references the PRD and task drafts against the original Idea and posts a concise review comment the admin uses in the review-chorus workflow.
Quick Start
Ask the agent to review the Chorus proposal with a given proposal UUID and post its verdict comment on that proposal.