Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate initial project planning documents (PVS, ADR, Tech Spec, Roadmap) from a project concept description. Use when starting a new project, when docs/planning/ contains placeholder files, or when user requests project planning document generation.
.claude/skills/aiskillstore-project-planning/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 7 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 205% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 54% | 0% |
Generate four essential planning documents to guide AI-assisted development. These documents maintain context coherence across coding sessions and prevent architectural drift.
docs/planning/ show "Awaiting Generation" status/plan command with project description| Document | Location | Purpose | |----------|----------|---------| | Project Vision & Scope | docs/planning/project-vision.md | What & Why - problem, solution, scope | | Technical Spec | docs/planning/tech-spec.md | How - architecture, data model, APIs | | Development Roadmap | docs/planning/roadmap.md | When - phased implementation plan | | Architecture Decision Record | docs/planning/adr/adr-001-*.md | Key technical decisions with rationale |
Before generating, collect:
pyproject.toml and existing codeGenerate documents sequentially, as later documents reference earlier ones:
After generating each document, use the zen-mcp-server consensus tool to get expert review:
Use mcp__zen__consensus with gemini-3-pro-preview to review:
"Review this [document type] for sufficiency to begin development.
Evaluate:
1. SPECIFICITY: Are requirements concrete enough to implement?
2. COMPLETENESS: Are all critical sections filled with project-specific content?
3. FEASIBILITY: Are timelines and technical choices realistic?
4. CLARITY: Can a developer understand what to build from this?
5. GAPS: What critical information is missing?
Respond with:
- READY: Document is sufficient to begin work
- NEEDS REVISION: [List specific improvements required]
Document content:
[paste document content]"Review each document in order:
If any document NEEDS REVISION, incorporate feedback and re-review before proceeding.
After all documents pass review:
Use template: templates/pvs-template.md
Focus on:
Use template: templates/adr-template.md
Create ADR for:
Format: adr-001-{decision-slug}.md
Use template: templates/tech-spec-template.md
Include:
Use template: templates/roadmap-template.md
Structure as:
Each phase needs:
When generating, incorporate known information:
python# From pyproject.toml / cookiecutter context python_version = "3.12" project_name = "Fragrance Rater" project_slug = "fragrance_rater" cli_framework = "Click" containerization = "Docker"
Before completing generation:
docs/planning/Templates are in templates/ directory:
pvs-template.md - Project Vision & Scope structureadr-template.md - Architecture Decision Record structuretech-spec-template.md - Technical Spec structureroadmap-template.md - Development Roadmap structureFor comprehensive documentation on each document type, see reference/ directory:
reference/document-guide.md - Full guidance for all document typesreference/prompting-patterns.md - How to use documents during developmentInstruct user to:
[ ]When user says: "I want to build a CLI tool for managing personal finances..."
templates/pvs-template.mddocs/planning/project-vision.md with finance CLI specificsmcp__zen__consensus with gemini-3-pro-preview → READY or revisetemplates/adr-template.mddocs/planning/adr/adr-001-database-choice.md for SQLite decisionmcp__zen__consensus with gemini-3-pro-preview → READY or revisetemplates/tech-spec-template.mddocs/planning/tech-spec.md with Python/Click/SQLite stackmcp__zen__consensus with gemini-3-pro-preview → READY or revisetemplates/roadmap-template.mddocs/planning/roadmap.md with phased implementationmcp__zen__consensus with gemini-3-pro-preview → READY or revisescripts/validate-planning-docs.pymcp__zen__consensus with gemini-3-pro-preview:
Review this Project Vision & Scope document for Fragrance Rater.
EVALUATION CRITERIA:
1. SPECIFICITY - Can a developer implement from these requirements?
2. COMPLETENESS - All sections filled with project-specific content?
3. FEASIBILITY - Realistic scope for the described constraints?
4. CLARITY - Unambiguous success criteria and scope boundaries?
5. GAPS - Any critical missing information?
RESPOND:
- READY: Sufficient to proceed to next document
- NEEDS REVISION: [Specific improvements with examples]
DOCUMENT:
[Full document content here]This skill requires the zen-mcp-server for consensus review. If not available, skip Step 3 and proceed with manual review.
Configuration: Ensure mcp__zen__consensus tool is accessible.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 51,024 | 18,191 | -64% | 1 | 1 | 0% | 7,247 | 2,415 | -67% | 0 | 0 | — |
case-02 | fail→fail | 38,416 | 39,062 | +2% | 1 | 1 | 0% | 6,630 | 10,620 | +60% | 0 | 0 | — |
case-03 | fail→fail | 51,332 | 17,387 | -66% | 1 | 1 | 0% | 8,263 | 2,621 | -68% | 0 | 0 | — |
case-04 | fail→pass | 11,996 | 28,211 | +135% | 1 | 1 | 0% | 1,949 | 5,947 | +205% | 0 | 0 | — |
case-05 | fail→pass | 48,370 | 42,765 | -12% | 1 | 1 | 0% | 7,309 | 9,309 | +27% | 0 | 0 | — |
case-06 | fail→fail | 11,186 | 36,866 | +230% | 1 | 1 | 0% | 888 | 6,680 | +652% | 0 | 0 | — |
case-07 | pass→fail | 23,257 | 38,880 | +67% | 1 | 1 | 0% | 3,667 | 5,486 | +50% | 0 | 0 | — |
case-08 | fail→pass | 30,563 | 24,385 | -20% | 1 | 1 | 0% | 5,007 | 6,490 | +30% | 0 | 0 | — |
case-09 | pass→pass | 22,242 | 22,398 | +1% | 1 | 1 | 0% | 3,507 | 4,734 | +35% | 0 | 0 | — |
case-10 | pass→pass | 45,984 | 33,243 | -28% | 1 | 1 | 0% | 8,217 | 6,631 | -19% | 0 | 0 | — |
case-11 | fail→pass | 39,558 | 22,753 | -42% | 1 | 1 | 0% | 6,136 | 4,805 | -22% | 0 | 0 | — |
case-12 | pass→fail | 20,561 | 82,867 | +303% | 1 | 1 | 0% | 2,994 | 10,378 | +247% | 0 | 0 | — |
case-13 | pass→pass | 25,641 | 14,738 | -43% | 1 | 1 | 0% | 3,246 | 4,075 | +26% | 0 | 0 | — |
case-14 | pass→pass | 41,690 | 43,334 | +4% | 1 | 1 | 0% | 6,588 | 8,646 | +31% | 0 | 0 | — |
case-15 | fail→pass | 47,358 | 34,995 | -26% | 1 | 1 | 0% | 3,431 | 5,295 | +54% | 0 | 0 | — |
case-16 | fail→pass | 34,412 | 44,167 | +28% | 1 | 1 | 0% | 5,720 | 6,235 | +9% | 0 | 0 | — |
case-17 | fail→fail | 18,828 | 13,569 | -28% | 1 | 1 | 0% | 2,137 | 3,308 | +55% | 0 | 0 | — |
case-18 | pass→fail | 21,374 | 15,717 | -26% | 1 | 1 | 0% | 3,439 | 2,191 | -36% | 0 | 0 | — |
case-19 | fail→pass | 15,878 | 14,033 | -12% | 1 | 1 | 0% | 1,704 | 3,593 | +111% | 0 | 0 | — |
case-20 | pass→pass | 16,203 | 17,331 | +7% | 1 | 1 | 0% | 1,823 | 3,820 | +110% | 0 | 0 | — |
case-21 | pass→fail | 21,903 | 15,931 | -27% | 1 | 1 | 0% | 3,215 | 2,092 | -35% | 0 | 0 | — |
case-22 | pass→pass | 11,301 | 25,849 | +129% | 1 | 1 | 0% | 1,777 | 4,141 | +133% | 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 15 counted toward the lift figure. The other 7 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 +14 percentage points is the difference between those two pass rates over the 15 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
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.