Triggers#

• When progressing a new project from start to finish
• When full lifecycle orchestration is needed

Usage#

# Full pipeline
/product  " Community feature development " 

 # Specify level (affects skip rules)
/product --level 3  " Community feature development " 

 # Start from a specific stage
/product --from specify  " Community feature development " 
 /product --from design  " Community feature development " 

 # Up to a specific stage only
/product --to epic  " Community feature development "

Parameters#

Options#

Execution Flow#

Step 1: Project Initialization#

1. Generate slug from project description (kebab-case)
2. Auto-determine BMAD project level
   • Level 0: Simple bug fix → Skip Discovery/Specification/Launch
   • Level 1: Small-scale feature addition → Skip Discovery/Specification
   • Level 2: Medium-scale feature → Full execution
   • Level 3: Large-scale feature/new module → Full execution
   • Level 4: New project → Full execution (deep)
3. Create .pipeline/{slug}.yaml state file
4. Verify/create docs/ directory

Step 2: Discovery (Analysis)#

Skip condition: Level 0-1

→ /product:discover  " {description} "

• cc-product: User interview, assumption validation, experiment design
• cc-product: Competitive analysis, vision formulation
• cc-product: /product:research, /product:brainstorm
• Artifacts: docs/discovery-{slug}.md

Step 3: Specification (Spec Verification)#

Skip condition: Level 0-1

→ /product:specify  " {description} "

• cc-spec-clarity: Socratic interview, Seed spec, ambiguity evaluation
• cc-spec-clarity: Contrarian review + Simplifier simplification (parallel)
• cc-spec-clarity: 3-stage evaluation (mechanical/semantic/consensus)
• Gate: Specification Gate (ambiguity ≤ 0.2, convergence ≥ 0.95, Contrarian 0, consensus 3/3)
• Artifacts: docs/seed-spec-{slug}.md, docs/socratic-discovery-{slug}.md, docs/decision-log-{slug}.md, docs/spec-evaluation-{slug}.md

Step 4: Planning#

→ /product:plan

• cc-product: /product:product-brief, /product:prd or /product:tech-spec
• cc-product: Business model
• cc-product Analyst: Requirements analysis, BDD Acceptance Criteria
• Gate: BMAD Analysis Gate
• Artifacts: PRD/Tech Spec + BDD Acceptance Criteria

Step 5: Design#

→ /product:design

• cc-product: /product:architecture, /product:create-ux-design
• cc-product Architect + UX Designer: Parallel review
• Gate: BMAD Solutioning Gate
• Artifacts: Architecture Doc + UX Spec

Step 5.5: Scaffold#

Skip condition: Level 0-1

→ /product:scaffold

• cc-bricks-feature: Mason brick scaffolding per feature (co-bricks)
• cc-flutter-dev: /bdd:generate for each feature + co-test-gen build_runner
• Gate: Scaffold Gate (brick output, .feature files, test gen, step stubs)
• Artifacts: Scaffolded feature directories, .feature files, widget + patrol test stubs

Step 6: Epic Creation#

→ /product:epic

• cc-dev-cycle: /zenhub:epic
• cc-product: /product:create-story, /product:sprint-planning
• cc-product PM: Story points, labeling
• Gate: BMAD Planning Gate
• Artifacts: ZenHub Epic + linked Stories

Step 7: Development#

→ /product:develop --epic {epic_number}

• Iterate through Stories within Epic, execute cc-dev-cycle 13 steps for each
• cc-product: /product:dev-story
• cc-product Flutter/Backend Dev: Implementation review
• Gate: BMAD Implementation Gate
• Artifacts: Implementation code + PR

Step 8: Launch#

Skip condition: Internal features, minor improvements

→ /product:launch

• cc-product: GTM strategy, ICP definition, messaging
• cc-product: A/B testing, cohort analysis
• Artifacts: docs/gtm-{slug}.md

Step 9: Completion#

• Final state file update
• Full pipeline summary report output
• Artifact list cleanup

Stage Transitions#

After each stage completion:

1. Verify artifact existence
2. Verify gate passage (if applicable)
3. Update state file
4. Request user confirmation before entering next stage

Error Handling#

• Gate failure: Display failure reason, guide to retry after fix
• Source plugin not installed: Warn and guide manual progression of that stage
• Interruption: Record last position in state file, resume later with --from option