Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Feature lifecycle: design, plan, implement, validate, release. Phase-gated workflow.
.claude/skills/notque-feature-lifecycle/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 62% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 8% | 0% |
Phase-gated feature workflow: DESIGN > PLAN > IMPLEMENT > VALIDATE > RELEASE. Each phase must pass its gate before the next begins.
Determine which phase to execute based on feature state:
.feature/ exists, check current phase:bash python3 ~/.claude/scripts/feature-state.py status Route to the phase indicated by the state machine.
Architecture Change Handoff targeting feature-lifecycle -> DESIGN; the design phase adopts it before approach workreferences/design.mdreferences/plan.mdreferences/implement.mdreferences/validate.md — run quality gate commands, do not reason about whether they would pass. Paste exit codes and output.references/release.mdreferences/pipeline.mdRead references/shared.md for directory structure, state management commands, context loading rules, and naming conventions. All state operations go through python3 ~/.claude/scripts/feature-state.py -- never manipulate state files directly.
DESIGN -> PLAN -> IMPLEMENT -> VALIDATE -> RELEASE
| | | | |
v v v v v
design.md plan.md impl.md report.md PR mergedEach phase produces an artifact consumed by the next. Skipping phases is not supported because downstream phases depend on artifacts from earlier phases.
| Signal | Load These Files | Why | |---|---|---| | Phase 1 DESIGN: idea to design document | design.md | Loads detailed guidance from design.md. | | Architecture Change Handoff targets feature-lifecycle | design.md | Validate and adopt architecture scope/interface/migration; create/register the feature ADR in DESIGN. | | errors, error handling | error-handling.md | Loads detailed guidance from error-handling.md. | | Phase 3 IMPLEMENT: dispatching task waves to domain agents | implement.md | Loads detailed guidance from implement.md. | | running the full design-to-release pipeline | pipeline.md | Loads detailed guidance from pipeline.md. | | Phase 2 PLAN: decomposing design into wave-ordered tasks | plan.md | Loads detailed guidance from plan.md. | | Phase 5 RELEASE: PR merge, tagging, worktree cleanup | release.md | Loads detailed guidance from release.md. | | feature state directory layout, naming, write protection | shared.md | Loads detailed guidance from shared.md. | | Phase 4 VALIDATE: running quality gates | validate.md | Loads detailed guidance from validate.md. |
| Error | Cause | Solution | |-------|-------|----------| | Phase mismatch | User requests phase N but state is at phase M | Report current state, suggest correct next phase | | Missing artifact | Previous phase did not produce expected output | Route back to previous phase | | Gate failure | Phase requirements not met | Report what failed, suggest fixes |
| Task Signal | Load | |-------------|------| | Phase routing (design/plan/implement/validate/release) | Phase reference (design.md, plan.md, implement.md, validate.md, release.md) | | End-to-end or full pipeline | references/pipeline.md | | State commands, directory structure, naming | references/shared.md | | Error, failure, blocked, gate fail, retry, deviation | references/error-handling.md |
references/design.md -- Design phase: explore requirements, discuss trade-offsreferences/plan.md -- Plan phase: break design into wave-ordered tasksreferences/implement.md -- Implement phase: dispatch tasks to domain agentsreferences/validate.md -- Validate phase: run quality gatesreferences/release.md -- Release phase: merge, tag, cleanupreferences/pipeline.md -- End-to-end orchestration across all phasesreferences/shared.md -- State conventions shared across all phasesreferences/error-handling.md -- Recovery patterns for phase errors, gate failures, agent dispatch failures| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→pass | 8,096 | 2,047 | -75% | 1 | 1 | 0% | 1,176 | 1,508 | +28% | 0 | 0 | — |
case-22 | pass→pass | 11,573 | 12,131 | +5% | 1 | 1 | 0% | 2,084 | 3,150 | +51% | 0 | 0 | — |
case-01 | fail→fail | 18,789 | 4,918 | -74% | 1 | 1 | 0% | 3,113 | 1,403 | -55% | 0 | 0 | — |
case-02 | fail→fail | 6,018 | 3,794 | -37% | 1 | 1 | 0% | 894 | 1,401 | +57% | 0 | 0 | — |
case-03 | fail→fail | 8,213 | 4,941 | -40% | 1 | 1 | 0% | 1,275 | 1,414 | +11% | 0 | 0 | — |
case-05 | fail→fail | 10,523 | 1,424 | -86% | 1 | 1 | 0% | 1,691 | 1,430 | -15% | 0 | 0 | — |
case-06 | fail→fail | 9,527 | 3,711 | -61% | 1 | 1 | 0% | 1,485 | 1,795 | +21% | 0 | 0 | — |
case-07 | fail→pass | 8,624 | 3,245 | -62% | 1 | 1 | 0% | 1,228 | 1,692 | +38% | 0 | 0 | — |
case-08 | fail→pass | 8,158 | 1,611 | -80% | 1 | 1 | 0% | 1,295 | 1,455 | +12% | 0 | 0 | — |
case-09 | fail→pass | 6,900 | 3,202 | -54% | 1 | 1 | 0% | 1,045 | 1,693 | +62% | 0 | 0 | — |
case-10 | fail→pass | 11,491 | 4,595 | -60% | 1 | 1 | 0% | 1,822 | 1,964 | +8% | 0 | 0 | — |
case-11 | pass→pass | 8,596 | 4,869 | -43% | 1 | 1 | 0% | 1,321 | 2,016 | +53% | 0 | 0 | — |
case-12 | pass→pass | 11,065 | 7,225 | -35% | 1 | 1 | 0% | 1,685 | 2,404 | +43% | 0 | 0 | — |
case-13 | fail→pass | 9,630 | 1,498 | -84% | 1 | 1 | 0% | 1,363 | 1,407 | +3% | 0 | 0 | — |
case-14 | fail→pass | 8,498 | 2,256 | -73% | 1 | 1 | 0% | 1,333 | 1,500 | +13% | 0 | 0 | — |
case-15 | fail→fail | 11,023 | 2,208 | -80% | 1 | 1 | 0% | 1,616 | 1,513 | -6% | 0 | 0 | — |
case-16 | pass→pass | 6,238 | 3,796 | -39% | 1 | 1 | 0% | 938 | 1,766 | +88% | 0 | 0 | — |
case-17 | pass→pass | 6,710 | 4,812 | -28% | 1 | 1 | 0% | 1,014 | 1,898 | +87% | 0 | 0 | — |
case-18 | pass→pass | 6,688 | 3,424 | -49% | 1 | 1 | 0% | 968 | 1,654 | +71% | 0 | 0 | — |
case-19 | pass→pass | 9,504 | 3,955 | -58% | 1 | 1 | 0% | 1,418 | 1,810 | +28% | 0 | 0 | — |
case-20 | pass→pass | 11,317 | 6,457 | -43% | 1 | 1 | 0% | 2,108 | 2,335 | +11% | 0 | 0 | — |
case-21 | pass→pass | 13,086 | 8,784 | -33% | 1 | 1 | 0% | 2,157 | 2,611 | +21% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +32 percentage points is the difference between those two pass rates over the 19 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.