mattgierhart/PRD-driven-context-engineering / .claude/skills/ghm-gate-check
ghm-gate-check
VerifiedValidates gate criteria before PRD lifecycle advancement by delegating to the readiness scoring pipeline (scripts/readiness.py). Returns a graduated PASS / WARN / BLOCK verdict with top blockers and their causal chain. Triggers before advancing from v0.X to v0.Y or explicit `/ghm-gate-check`.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.3 KB |
| assets/gate-checklist-template.md | markdown | 7.3 KB |
| references/examples.md | markdown | 9.4 KB |
| references/gate-criteria.md | markdown | 12.3 KB |
references/examples.md
View on GitHub# Gate Check Examples
**Purpose**: Good and bad patterns for gate validation with pass/fail scenarios.
> **Note**: As of `ghm-gate-check@evolved-2026-04-17`, verdicts come from `scripts/readiness.py` — a graduated score (0–100) across three layers. The examples below show the JSON-driven report shape; the "Pass/Fail Pattern Summary" at the end covers the *content* patterns (what makes a CFD entry good, what makes a risk actionable).
---
## Graduated-Score Report Example (BLOCK)
**Context**: OriginStamp, evaluating v0.7 → v0.8 transition. `readiness.py run` exits 2.
```markdown
## Gate Check Report: v0.7 → v0.8 (Build → Deployment)
**Verdict**: BLOCK
**Stage Score**: 36.6 / 100 (warn < 70, block < 50)
**Date**: 2026-04-17
### Stage Dimensions
| Dimension | Score | Weight |
|-----------|-------|--------|
| required_ids_present | 50.0 | 0.300 |
| relevant_sot_readiness | 55.4 | 0.300 |
| cross_ref_integrity | 95.6 | 0.200 |
| downstream_epic_readiness | 54.3 | 0.200 |
### Top Blockers (leverage view)
1. **SoT/SoT.TESTING.md** (score 0.0) — blocks 8 EPICs: EPIC-01..EPIC-08 — impact 800
2. **SoT/SoT.API_CONTRACTS.md** (score 55.8) — blocks 7 EPICs: EPIC-02..EPIC-08 — impact 309
3. **SoT/SoT.USER_JOURNEYS.md** (score 0.0) — blocks 1 EPIC: EPIC-07 — impact 100
### Unmet Criteria (high severity first)
- [high] TEST: Found 0 TEST- entries; gate requires ≥1
- [high] SoT/SoT.TESTING.md scores 0.0 — drags down stage readiness
- [high] 7 dangling ID reference(s) across repo: TEST-008, UJ-001, UJ-002, …
### Recommendation
Do not advance. Address top blockers in order — fixing the highest-impact SoT file cascades up the graph.
**Next action**: Populate SoT.TESTING.md with TEST- entries covering every API-/BR- in scope. Unblocks all 8 EPICs at once.
```
The power here is in the leverage view: 8 EPICs at WARN suggests 8 problems; the connected scorer reveals 1 root cause (empty `SoT.TESTING.md`). That's the fix-first action.
---
## Graduated-Score Report Example (PASS)
**Context**: Hypothetical mature repo advancing v0.6 → v0.7. Score 84.
```markdown
## Gate Check Report: v0.6 → v0.7 (Architecture → Build)
**Verdict**: PASS
**Stage Score**: 84.0 / 100 (warn < 70, block < 50)
### Stage Dimensions
| Dimension | Score | Weight |
|-----------|-------|--------|
| required_ids_present | 100.0 | 0.375 |
| relevant_sot_readiness | 82.3 | 0.375 |
| cross_ref_integrity | 98.0 | 0.250 |
| downstream_epic_readiness | n/a | — |
### Top Blockers
None — all SoT files passing.
### Recommendation
Advance to v0.7. Run `ghm-status-sync` to update the README dashboard.
```
---
## Content Patterns — Good vs Bad Artifacts
The scorer enforces structure (IDs exist, files are populated, refs resolve). The patterns below cover *content quality* — what makes individual artifacts good, which is still a human judgment the scorer can't fully automate.
---
## Good Example: v0.2 → v0.3 Gate Pass
### Context
Project management tool for design teams
### Artifacts Submitted
**Competitive Analysis (CFD)**:
- CFD-021: Figma (strengths: design integration, weaknesses: weak project management)
- CFD-022: Asana (strengths: robust PM, weaknesses: no design tools)
- CFD-023: Notion (strengths: flexible, weaknesses: not design-specific)
**Product Type (BR-012)**:
- Classification: Unbundle
- Rationale: Extract project management from generic tools, specialize for designers
**Positioning**:
- Differentiation: Built-in asset management, Figma integration, design workflows
- Target: Design teams switching from Asana (need design-specific features)
### Validation Result: ✅ PASS
**Why**:
- ✅ 3 competitors analyzed thoroughly
- ✅ Product type justified (unbundle strategy clear)
- ✅ Differentiation specific (not "better PM")
- ✅ Clear switching target (Asana users needing design features)
**Action**: Advance to v0.3 (Commercial Model)
---
## Bad Example 1: v0.1 → v0.2 Gate Blocked
### Context
AI-powered scheduling assistant
### Artifacts Submitted
**CFD Entries**:
- CFD-001: "Users want better scheduling" (no evidence, no specifics)
- CFD-002: "Calendars are annoying" (vague, no quantification)
**Problem Statement**:
"People hate scheduling meetings"
**Value Hypothesis**:
"Our AI will make it easier"
### Validation Result: ❌ BLOCKED
**Blockers**:
1. ❌ No evidence for pain points (CFD entries empty)
2. ❌ Problem too vague ("hate scheduling" - what specifically?)
3. ❌ Value unclear ("easier" - how? by how much?)
**Required Actions**:
- [ ] Conduct 10 user interviews (find specific pain: time zone confusion? back-and-forth emails?)
- [ ] Quantify problem (how many hours/week lost? how much frustration?)
- [ ] Define specific value ("reduce scheduling time from 5 min to 30 sec")
**Re-check Date**: 2 weeks (after user research)
---
## Bad Example 2: v0.3 → v0.4 Conditional Pass
### Context
CRM for real estate agents
### Artifacts Submitted
**Pricing Model (BR-034)**:
- Model: Tiered pricing
- Tiers: Free, Pro ($50/mo), Enterprise (custom)
- Justification: "Feels right, competitors charge $30-100"
**Success Metrics (KPI-011, KPI-012)**:
- Primary: 1,000 total signups Month 1
- Secondary: 100 Pro subscribers
**Moat Analysis**:
- Defensibility: "We'll build integrations with MLS systems"
- Competitor moats: "They're slow, we're fast"
**Features (FEA-XXX)**:
- 25 features listed, all marked "high priority"
### Validation Result: ⚠️ CONDITIONAL PASS
**Issues**:
1. ⚠️ Pricing not validated (survey? willingness to pay?)
2. ⚠️ Primary KPI is vanity metric (total signups, not activation)
3. ⚠️ Moat weak ("integrations" - competitors can copy)
4. ⚠️ No feature prioritization (can't build 25 features for MVP)
**Conditions to Address (within 1 week)**:
- [ ] Survey 20 real estate agents: "Would you pay $50/mo for [value prop]?"
- [ ] Change primary KPI to "activated users" (define activation)
- [ ] Re-assess moat (data moat? network effects? brand?)
- [ ] Prioritize features: Mark 5 as MVP, rest as "post-MVP"
**Decision**: Can advance, but must fix conditions before v0.5
---
## Good Example: v0.7 → v0.8 Gate Pass
### Context
Dark mode feature for SaaS product
### Artifacts Submitted
**EPICs**:
- EPIC-012: CSS theme variables (1 week)
- EPIC-013: User preference storage (3 days)
- EPIC-014: Toggle UI component (2 days)
**Tests (TEST-XXX)**:
- TEST-034: Dark mode toggle saves preference
- TEST-035: Theme applies correctly across all screens
- TEST-036: Accessibility (WCAG AA compliance)
- Coverage: 85%
- All tests passing ✅
**Code Traceability**:
- `ThemeProvider.tsx`: @implements BR-089 (User preference persistence)
- `ToggleComponent.tsx`: @implements UI-012 (Dark mode toggle)
### Validation Result: ✅ PASS
**Why**:
- ✅ EPICs right-sized (days to 1 week, fit in context)
- ✅ Comprehensive test coverage (85%, includes accessibility)
- ✅ All tests passing
- ✅ Code traceable (can find which code implements which spec)
**Action**: Advance to v0.8 (Deployment planning)
---
## Bad Example 3: v0.5 → v0.6 Gate Blocked
### Context
Marketplace for freelance designers
### Artifacts Submitted
**Risks (RISK-XXX)**:
- RISK-001: "Might not work" (no specifics)
- RISK-002: "Competition" (too vague)
**Tech Stack (TECH-XXX)**:
- TECH-001: "Build custom payment processor" (buy decision: none)
- TECH-002: "Build custom search engine" (buy decision: none)
- TECH-003: "Build custom email system" (buy decision: none)
### Validation Result: ❌ BLOCKED
**Blockers**:
1. ❌ Risks too vague (not actionable)
2. ❌ Build vs buy: Building everything from scratch (NIH syndrome)
3. ❌ No risk mitigations
**Required Actions**:
- [ ] Define specific risks (e.g., "Cold start problem: need both designers and clients to launch")
- [ ] Mitigation for each risk (e.g., "Recruit 50 designers pre-launch")
- [ ] Reconsider build vs buy (Stripe for payments, Algolia for search, SendGrid for email)
- [ ] Justify any build decision (why not buy? ROI? Strategic value?)
**Re-check Date**: 1 week
---
## Pass/Fail Pattern Summary
### Patterns That Pass
**Good CFD Entries**:
- Specific pain point
- Evidence (quote, survey data, support ticket count)
- Quantified (frequency, severity)
**Good KPIs**:
- Actionable (not vanity)
- Thresholds (success/caution/failure)
- Tied to business outcome
**Good Features**:
- Prioritized (MVP vs post-MVP)
- Tied to business rules and outcomes
- Traceability (FEA → BR → KPI)
**Good Tests**:
- > 80% coverage
- All passing
- Include edge cases, accessibility
---
### Patterns That Block
**Bad CFD Entries**:
- Vague ("users want better tools")
- No evidence (hunches)
- Not quantified
**Bad KPIs**:
- Vanity metrics (total signups)
- No thresholds (how do we know if we're succeeding?)
- Not tied to revenue/retention
**Bad Features**:
- All "high priority" (no trade-offs)
- No traceability (can't connect to business value)
- Scope creep (50 features for MVP)
**Bad Build Decisions**:
- NIH syndrome (build everything)
- No justification (why not Stripe? Twilio? AWS?)
- Ignoring opportunity cost
---
## Conditional Pass Criteria
**When to Use**:
- 80%+ of criteria met
- Issues are fixable in < 1 week
- Blocker conditions not present
**Example Conditions**:
- Pricing model needs validation (survey 20 users)
- KPI needs to change from vanity to actionable
- Features need prioritization (mark MVP vs post-MVP)
**Follow-up**:
- Set deadline (1 week)
- Assign owner
- Re-validate when complete
---
*Reference: Use these examples when performing gate checks in `ghm-gate-check` skill.*
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |