Overview
The greenfield workflow:- Initialize — Set up hatch3r
- Generate specs — Use
project-specto create documentation - Create roadmap — Use
roadmapto plan phases - Fill board — Generate GitHub issues with
board-fill - Implement — Use
board-pickupto build features
1. Initialize Your Project
Start with an empty repository or create a new one:Select tools
Choose which coding tools you use:
- Cursor
- GitHub Copilot
- Claude Code
- OpenCode
- Windsurf
- Amp
- Codex CLI
- Gemini CLI
- Cline / Roo Code
2. Generate Project Specs
Run theproject-spec command to generate complete documentation:
What project-spec Does
project-spec spawns 7 parallel researcher sub-agents:Stack Researcher
Recommends languages, frameworks, databases, and infrastructure calibrated to your company stage
Feature Researcher
Breaks your vision into features organized by domain/module with priority ordering
Architecture Researcher
Designs system architecture, data flows, and integration points
Pitfalls Researcher
Identifies risks, anti-patterns, security concerns, and scalability bottlenecks
UX Researcher
Analyzes user journeys, accessibility requirements, and visual design direction
Business Model Researcher
Validates business model, researches competitors, and recommends GTM strategy
Production Researcher
Designs infrastructure, scaling strategy, observability, and cost modeling
Running project-spec
Answer discovery questions
project-spec will ask about:Core Vision:
- Project name
- Vision and purpose (one paragraph)
- Target platforms (web, mobile, CLI, API, etc.)
- Key user personas and goals
- Known constraints (budget, timeline, tech mandates)
- Company stage (pre-revenue, early-revenue, growth, scale, enterprise)
- Team composition and size
- Current user/revenue scale
- Funding/runway status
- Regulatory/compliance needs
- Business model type (SaaS, marketplace, platform, etc.)
- Revenue model (subscription, transactional, freemium, etc.)
- Top 3 competitors
- Target market segments / ICP
- Key business metrics/KPIs
Review researcher outputs
project-spec presents a merged summary of all researcher findings:
- Stack recommendations with trade-offs
- Feature catalog with complexity estimates
- Architecture overview and decisions
- Risk register (technical and business)
- UX analysis and competitive research
- Market sizing and GTM recommendations
- Infrastructure blueprint and cost model
Confirm specs
Review and adjust the generated specs before writing files:Technical specs (
docs/specs/technical/):00_glossary.md— Stable IDs for entities, events, modules01_overview.md— Technical overview02_{module}.md— One spec per module
docs/specs/business/):00_business_glossary.md— Business entities and metrics01_business_overview.md— Business model and market context02_{domain}.md— One spec per business domain03_competitive_analysis.md— Competitor research04_gtm_strategy.md— Go-to-market plan05_production_blueprint.md— Infrastructure and cost model
Generate ADRs
project-spec creates Architecture Decision Records (ADRs) for each significant decision:
- Framework/language choices
- Database selection
- Architecture patterns
- Authentication approach
- Payment/billing architecture
Create todo.md
project-spec generates an initial
todo.md with prioritized items from:- Feature researcher’s priority ordering
- Pitfalls researcher’s risk register
- Business model researcher’s GTM requirements
- Production researcher’s infrastructure needs
Example Output Structure
3. Create a Roadmap
Runroadmap to generate a phased plan:
Review phases
roadmap breaks work into dependency-ordered phases:
- Phase 1: Foundation — Core infrastructure and architecture
- Phase 2: MVP Features — Minimum viable product
- Phase 3: Growth — Scaling and optimization
- Phase 4: Polish — UX improvements and nice-to-haves
Identify parallel lanes
roadmap identifies work that can proceed in parallel:
- Frontend and backend can often progress independently
- Infrastructure setup can run alongside feature development
- Documentation can be written concurrently
4. Fill the Board
Runboard-fill to create GitHub issues:
Triage items
board-fill asks clarifying questions about each item:
- Scope and boundaries
- Value and motivation
- Open questions or unknowns
- External dependencies
- Decomposition needs
5. Implementation Workflow
Pick Up First Issue
Runboard-pickup to start implementing:
Development Cycle
Implement feature
Delegates to the appropriate skill:
feature— End-to-end feature implementationbug-fix— Minimal fix with regression testrefactor— Code quality improvement
Best Practices
Start with MVP scope
Start with MVP scope
Focus on the smallest viable product first. Use the company stage assessment to calibrate recommendations:
- Pre-revenue: MVP-focused, lean analysis
- Growth: Scaling-aware, performance considerations
- Enterprise: Production-hardened, compliance-ready
Validate assumptions early
Validate assumptions early
Use the Business Model Researcher’s competitive analysis to validate your assumptions before building.
Document architectural decisions
Document architectural decisions
ADRs capture the context and rationale for decisions. Review and update them as you learn.
Iterate on specs
Iterate on specs
Specs are living documents. Update them as requirements change or implementation reveals new constraints.
Use stable IDs
Use stable IDs
Reference entities, events, and modules by their stable IDs from the glossary across docs, issues, and code.
Next Steps
Workflow Guide
Full project lifecycle from init to release
Board Fill
Learn how board-fill creates issues with dependency analysis
Documentation Structure
Understanding docs/specs/ and docs/adr/
Customization
Tailor hatch3r to your project needs

