PRD Quality Rubric & Anti-Pattern Detection (block placeholders)

[DevVig]

Summary
Define a rubric to score PRD quality (coverage, clarity, SMART goals, persona depth, NFR completeness) and detect anti-patterns (e.g., generic sample text, placeholder epics like "Authentication"/"Dashboard" without goals). Use this to warn or block further steps until addressed.

Why

• Forces alignment to goals/aims and prevents shallow PRDs from slipping into build

Scope

• Rubric in a2a/roles/analyst_prd_validator.py and patterns list in .a2dev/policies/assess.yaml

Tasks

• Define rubric criteria with weights (min pass threshold)
• Anti-pattern rules: placeholder headings, missing goals, missing personas, no KPIs, no NFRs, no rollout
• Add friendly remediation hints with links to wizard sections
• Include results in docs/analyst/prd-review.md and status line

Acceptance Criteria

• Cyon sample PRD is flagged with specific anti-patterns
• Users see concrete guidance to fix PRD via wizard or ingest

References

• Cyon docs/PRD.md