y-score-readinesslisted
Install: claude install-skill vignesh-s7/product-management
# y-score-readiness
The PO gate that runs before `prd-to-sdlc` produces any artifacts. Refuses to proceed if score < 70.
## When to trigger
- `/y-score <prd.md>`
- User says "is this ready to launch?", "readiness check", "score this PRD"
- Auto-invoked by `prd-to-sdlc` orchestrator stage 1
## The 7 dimensions
| # | Dimension | Weight | Anchor for full marks |
|---|---|---|---|
| 1 | Problem clarity | 15 | One-sentence statement, measurable, user-visible |
| 2 | Target user | 10 | Named persona + JTBD + frequency |
| 3 | Success metric | 20 | Leading + lagging metric with baseline + target |
| 4 | Constraints | 10 | Latency / cost / data residency / SLA defined |
| 5 | Compliance posture | 15 | Domain checklist done (HIPAA / PCI / GDPR / EU-AI-Act) |
| 6 | Eval plan | 20 | Golden set + judge + regression suite + drift monitor |
| 7 | Rollback plan | 10 | Kill switch + comms + data-recovery RTO defined |
## Scoring
- Per dimension: 0 (missing) / 50 (partial) / 100 (full)
- Weighted total = sum(dim_score × weight) / 100
- **Gate:** total ≥ 70 → proceed. < 70 → return findings, block downstream stages.
## Output format
```json
{
"score": 78,
"verdict": "go",
"dimensions": {
"problem_clarity": {"score": 100, "note": "..."},
"target_user": {"score": 50, "note": "Persona named but JTBD missing"},
...
},
"blockers": [],
"recommendations": ["Add LLM-as-judge calibration step to eval plan"]
}
```
## Domain-aware compliance dimension
Compliance dimens