AgentSkillsCN

project-plan

根据项目规格与调研结果,制定包含产品需求文档与史诗分解的完整项目计划。

SKILL.md
--- frontmatter
name: project-plan
description: Create comprehensive project plan with PRD and epic breakdown based on project specification and research
disable-model-invocation: false

description: Create comprehensive project plan with PRD and epic breakdown based on project specification and research.

The user input to you can be provided directly by the agent or as a command argument - you MUST consider it before proceeding with the prompt (if not empty).

User input:

$ARGUMENTS

Generate a Product Requirements Document (PRD) and identify epics based on project specification and upstream artifacts.

Research Approach: If planning needs market/business research, uses just-in-time research pattern (.cursor/skills/just-in-time-research/SKILL.md)

Subagent Parallelization

This command benefits from parallel execution:

Epic Spec Drafting - Spawn parallel speck-scribe:

code
├── [Parallel] speck-scribe: Draft E001 epic.md placeholder
├── [Parallel] speck-scribe: Draft E002 epic.md placeholder
├── [Parallel] speck-scribe: Draft E003 epic.md placeholder
└── [Wait] → Create all epic directories with drafted specs

Research Phase (if needed) - Spawn parallel speck-researcher:

code
├── [Parallel] speck-researcher: "Market sizing for [domain]"
├── [Parallel] speck-researcher: "Competitor analysis"
├── [Parallel] speck-researcher: "Pricing strategies"
└── [Wait] → Embed findings in PRD

Speedup: Nx (where N = number of epics) for spec drafting.

  1. Load project context and foundation artifacts:

    • Find active project directory (check cwd, then scan specs/projects/)
    • Load and validate project.md
    • Check for existing research reports: -research-report-.md (from earlier commands)
    • Check for brownfield artifacts:
      • project-landscape-overview.md (brownfield code analysis)
      • project-import.md (brownfield non-code aspects)
    • Load REQUIRED foundation artifacts:
      • architecture.md (from /project-architecture) - REQUIRED! Architecture decisions inform planning
      • ux-strategy.md (from /project-ux) - for UX principles and user journeys
      • context.md (from /project-context) - for constraints and requirements
    • Load OPTIONAL foundation artifacts:
      • domain-model.md (from /project-domain) - for domain terminology, rules, and principles
      • design-system.md (from /project-design-system) - for UI requirements
      • constitution.md (from /project-constitution) - for technical principles
    • If project.md missing: ERROR "Run /project-specify first"
    • If architecture.md missing: ERROR "Run /project-architecture first - design decisions must inform planning!"

    Check for Active Recipe:

    • Look for _active_recipe: in project.md metadata
    • If found, load .speck/recipes/[recipe-name]/recipe.yaml
    • Recipe provides:
      • suggested_epics: → Pre-defined epic structure for this project type
      • external_services: → Recommended services to consider
      • patterns: → Implementation patterns to reference
    • Use recipe's suggested_epics as starting point in Phase 2 (Epic Identification)
    • Customize epic list based on project-specific requirements

    Note: UX-strategy.md and context.md are strongly recommended. Design-system.md is optional but valuable for UI-heavy projects.

  2. Determine project scale (Level 0-4):

    • Analyze project scope and complexity
    • Consider: feature count, user types, integrations, technical complexity
    • Map to scale:
      • Level 0: Single atomic change
      • Level 1: 1-10 stories, 1 epic
      • Level 2: 5-15 stories, 1-2 epics
      • Level 3: 12-40 stories, 2-5 epics
      • Level 4: 40+ stories, 5+ epics
  3. Just-In-Time Research (if needed for planning):

    If planning requires market/business insights not yet available:

    • Market Sizing: Web search for market data, growth trends
    • Competition: Web search for competitor analysis
    • Business Model: Web search for pricing strategies, monetization patterns
    • Deep Research (if needed): Detailed market analysis, business model validation

    Document findings in PRD's "Research Informing This Plan" section.

  4. Planning execution phases:

    Phase 1: PRD Generation

    • Use .speck/templates/project/prd-template.md
    • Fill sections with project-specific content
    • Incorporate architecture decisions (REQUIRED):
      • Technology stack choices and rationale
      • Architectural patterns and constraints
      • Component structure and boundaries
      • Integration approaches
      • Performance and scalability considerations
    • Incorporate UX strategy:
      • UX principles and user journeys from ux-strategy.md
      • Design philosophy and patterns
    • Incorporate context constraints:
      • Non-functional requirements from context.md
      • Team skills and infrastructure constraints
      • Compliance and regulatory requirements
    • Incorporate design system (if exists):
      • UI component inventory
      • Design token requirements
      • Accessibility standards
    • Incorporate technical principles (if exists):
      • Constitution-mandated approaches from constitution.md
    • Incorporate domain expertise (if exists):
      • Use domain-model.md glossary terms in PRD consistently
      • Reference domain entities when defining features
      • Ensure domain rules/invariants are respected in requirements
      • Align feature scope with domain principles
    • Incorporate research findings if available
    • Include landscape-overview insights if brownfield
    • Ensure alignment with project.md vision
    • Mark sections as "[To be defined]" only if critical inputs genuinely missing

    Phase 2: Epic Identification

    If Active Recipe exists:

    • Start with recipe's suggested_epics: as base structure
    • E000 (Infrastructure) is already in recipe → use as first epic
    • Customize epics marked with [CUSTOMIZE] for project-specific features
    • Add/remove epics based on project.md scope
    • Skip optional epics (optional: true) unless project needs them

    If No Recipe:

    • Based on PRD scope, identify logical epic boundaries

    For all epics:

    • Each epic should:
      • Deliver standalone value
      • Be independently deployable
      • Have clear success criteria
      • Support 3-15 stories typically
    • Consider dependencies between epics
    • If domain-model.md exists:
      • Consider organizing epics around domain concepts
      • Use domain terminology in epic names and descriptions
      • Ensure each epic respects domain invariants
    • Consider business value for prioritization:
      • Revenue impact (enables monetization, unlocks paid tier)
      • Customer acquisition (improves funnel, reduces CAC)
      • Retention impact (reduces churn, increases LTV)
      • Competitive necessity (table stakes vs differentiation)
      • Market timing (time-sensitive opportunities)
    • Prioritize for MVP vs post-MVP based on:
      • User value + Business value + Technical dependency
      • Highest business ROI + foundational capabilities first

    Phase 2.5: Foundation Check (IMPORTANT)

    Before finalizing epics, check if a Developer Infrastructure epic is needed:

    Ask the user:

    code
    🏗️ Foundation Check
    
    Should I include "E000: Developer Infrastructure" epic as the first epic?
    
    This epic sets up:
    - Testing framework (unit, integration tests)
    - CI/CD pipeline (lint, test, build, deploy)
    - Linting & formatting (consistent code style)
    - Error tracking (Sentry or equivalent)
    - Environment configuration (.env patterns)
    
    Recommended: YES for any production-bound project
    Skip only if: Existing infrastructure or throwaway prototype
    
    Include Developer Infrastructure epic? [Y/n]
    

    If YES (default):

    • Add E000 epic before all other epics
    • Use tech stack from architecture.md to specify testing framework
    • Reference recipe (if used) for tech-specific tooling recommendations

    If NO:

    • Document skip reason in epics.md
    • Note: User accepts responsibility for infrastructure setup

    Phase 3: Epic Documentation

    • Create epics.md using .speck/templates/project/epics-list-template.md
    • For each epic include:
      • Epic ID and name
      • Business value statement (user value + business impact)
      • High-level scope
      • Success criteria
      • Dependencies
      • Estimated story count
      • Business metrics (if applicable): Revenue impact, CAC/LTV impact, retention impact
  5. PRD structure (adaptive by level):

    Level 0-1: Focused PRD

    • Problem & solution (1 page)
    • Single epic with stories
    • Basic requirements list
    • Simple success metrics

    Level 2: Standard PRD

    • Full problem space analysis
    • 1-2 epics with clear boundaries
    • Detailed requirements by epic
    • Phased delivery plan

    Level 3-4: Comprehensive PRD

    • Market & competitive analysis
    • 3-5+ epics with dependencies
    • Detailed user journeys
    • Architecture considerations
    • Rollout and adoption strategy
  6. Epic identification patterns:

    By User Journey

    • Onboarding epic
    • Core workflow epic
    • Admin/settings epic

    By Technical Layer

    • Infrastructure epic
    • API/backend epic
    • UI/frontend epic

    By Feature Area

    • Authentication epic
    • Data management epic
    • Reporting epic

    By Release Phase

    • MVP epic
    • Enhancement epic
    • Scale epic
  7. Generate planning artifacts:

    code
    [PROJECT_DIR]/
    ├── project.md (already exists)
    ├── PRD.md (new - with embedded research if any)
    ├── epics.md (new)
    └── epics/
        ├── E001-[epic-name]/
        ├── E002-[epic-name]/
        └── .../
    
  8. Epic directory initialization:

    • Create directory for each epic
    • Add placeholder epic.md with:
      • Epic name and ID
      • Initial scope from PRD
      • Dependencies identified
      • Placeholder sections for details
    • Ready for /epic-specify to enhance
  9. Validation and review:

    • PRD addresses all project goals
    • Epics cover full scope
    • No gaps or overlaps
    • Dependencies identified
    • Success metrics clear
  10. Output summary:

code
✅ Project Planning Complete!

Project: [Name]
Scale: Level [0-4]

Documents Generated:
- PRD.md (Product Requirements Document with embedded research)
- epics.md (Epic breakdown)

Epic Summary:
1. [Epic Name]: ~[X] stories
2. [Epic Name]: ~[X] stories
[...]

Total Estimated Scope: ~[X] stories

Next Steps:
1. Review PRD with stakeholders
2. Validate epic boundaries
3. Optional: /project-roadmap (for epic timeline and dependencies)
   - If you intentionally skipped /project-design-system and later decide you need it, run /project-design-system and then update PRD.md accordingly (or re-run /project-plan to re-synthesize PRD + epics with design-system inputs).
4. Quality gates:
   - /project-analyze (deep quality analysis)
   - /project-validate (go/no-go decision)
5. Begin epic development:
   - cd epics/E001-[epic-name]
   - /epic-specify (enhances the placeholder epic.md)
   - /epic-clarify (if needed)
   - /epic-plan (generates tech spec)
   
Note: Epic placeholder files are created with basic structure.
Running /epic-specify in each epic directory will complete the specification.
  1. Important notes:
    • For Level 0-1: Can skip epic level, go directly to story
    • For Level 3-4: Epic planning is critical for success
    • PRD is living document - update as learning occurs
    • Epic boundaries can be adjusted based on implementation learning