Agent skill
product
Product workflows router. Use when validating product ideas, writing specs, analyzing competitors, or tracking metrics. Triggers on "PRD", "competitive analysis", "metrics", "OKRs", "cohort", "requirements", "tech debt", "game-changing features", market positioning, retention analysis, or product strategy tasks.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/product
SKILL.md
Product Skill
Router skill for product management workflows. Detects task type and routes to appropriate workflow.
When This Skill Activates
You're working on:
- Competitive analysis or market positioning
- Product requirements documentation
- Goal setting and metrics definition
- User research synthesis
- Cohort and retention analysis
- PRD quality evaluation
- Technical debt prioritization
- Game-changing feature ideation
Pre-Work Spec Gate
Before proceeding, score the incoming request (5 dimensions, 0-2 each, max 10):
| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Problem | "Make it better" | "Improve onboarding" | "Reduce signup dropout by 20%" |
| Audience | None stated | "Our users" | "B2B SaaS teams >50 employees" |
| Constraints | None stated | "This quarter" | "No new hires, <$10k budget" |
| Success | None stated | "More engagement" | "NPS +10, churn <5%" |
| Done | Implied | "When shipped" | "Customer interviews confirm" |
Decision:
- ≥8: Proceed with workflow
- 5-7: Colleague mode — confirm understanding before analysis
- <5: Run
/hope:intentfirst — spec too vague
See fit-decision.md for full scoring guidance.
Workflow Selection
Announce which workflow you're using:
| Task Type | Workflow | Reference |
|---|---|---|
| Competitive analysis, market gaps, win/loss | Competitive Analysis | references/compete.md |
| Feature specs, requirements, MVP scoping | PRD Generation | references/prd.md |
| Goals, OKRs, KPIs, tracking systems | Metrics & Goals | references/metrics.md |
| User interviews, qualitative data, insights | Research Synthesis | references/research.md |
| Retention, LTV, churn, cohort data | Cohort Analysis | references/cohort.md |
| PRD review, completeness check, scoring | PRD Evaluation | references/prd-eval.md |
| Tech debt triage, remediation planning | Debt Prioritization | references/debt.md |
| Breakthrough features, innovation ideation | Game-Changing Features | references/game-changing-features.md |
| TAM/SAM/SOM, market opportunity sizing | Market Sizing | references/market-sizing.md |
| A/B tests, controlled experiments | Experimentation | references/experimentation.md |
Related Thinking Tools
From hope/skills/soul/references/tools/:
| Tool | When to Use |
|---|---|
| Second-Order Thinking | Predict consequences of product decisions |
| Pre-Mortem | Anticipate launch failures |
| Impact-Effort | Focus on high-impact, low-effort features |
From this skill's references/:
| Tool | When to Use |
|---|---|
| Jobs To Be Done | Understand what customers hire products for |
| Systems Archetypes | Recognize "limits to growth", "shifting burden" patterns |
Usage
- Detect which workflow applies based on user's task
- Announce: "I'm using the product skill for [workflow]"
- Load the appropriate reference file
- Execute the workflow with confirmation gates (see below)
Confirmation Gates
Multi-step workflows pause at checkpoints to prevent wasted work when intent drifts.
Gate Points:
| Phase | Gate |
|---|---|
| After research/discovery | ⚠️ CHECKPOINT: "Does this understanding match your intent?" |
| After approach selection | ⚠️ CHECKPOINT: "Here's my proposed approach. Should I proceed?" |
| Before final artifact | ⚠️ CHECKPOINT: "Ready to generate final output. Confirm?" |
Skip gates: Say "proceed without confirmation" to run uninterrupted.
In workflows: Each reference file should pause at these points:
### Phase 2: Analysis
[... phase content ...]
⚠️ **CHECKPOINT**: Present findings summary. Ask: "Does this capture the key insights? Any adjustments before I continue?"
Rules
- Use Ask tool to gather input before proceeding
- Ground recommendations in evidence
- Use story points, never time estimates
- Challenge assumptions and cut scope by default
- Quality footer required after each workflow completes — see quality-footer.md
Boundary
Product intuition is yours. Claude structures, not strategizes.
- Analysis surfaces patterns in data; user decides meaning
- Recommendations are hypotheses, not directives
- Market insights are inferences, not facts — always confidence-tagged
Product skill organizes thinking. Business judgment remains human.
Didn't find tool you were looking for?