AgentSkillsCN

project-clarify

通过有针对性的问答环节,识别项目规格中尚未明确的部分,进而判断哪些内容需进一步深入研究,而不仅仅是简单澄清。

SKILL.md
--- frontmatter
name: project-clarify
description: Identify underspecified areas in the project spec through strategic Q&A, then determine what needs deeper research beyond simple clarification
disable-model-invocation: false

description: Identify underspecified areas in the project spec through strategic Q&A, then determine what needs deeper research beyond simple clarification.

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

Goal: Detect and reduce ambiguity through targeted Q&A, identify areas requiring deeper research, and prepare the project for comprehensive planning.

Note: This clarification workflow is expected to run (and be completed) BEFORE invoking /project-plan. If the user explicitly states they are skipping clarification (e.g., proof of concept), you may proceed, but must warn that downstream rework risk increases.

Execution steps:

  1. Determine active project directory:

    • Check current working directory for project.md
    • If not found, scan specs/projects/ for most recent project
    • Parse PROJECT_DIR and PROJECT_SPEC paths
    • If no project found: ERROR "No active project found. Run /project-specify first"
  2. Load ONLY upstream context (do NOT load downstream artifacts):

    ALWAYS Load (upstream/input to clarify):

    • project.md (the spec to clarify)
    • project-import.md (if exists - brownfield non-code extraction)
    • project-landscape-overview.md (if exists - brownfield code extraction)

    NEVER Load (downstream/created AFTER clarify):

    • PRD.md - Created by /project-plan (comes AFTER clarify)
    • context.md - Created by /project-context (comes AFTER clarify)
    • architecture.md - Created by /project-architecture (comes AFTER clarify)
    • design-system.md - Created later (comes AFTER clarify)
    • ux-strategy.md - Created by /project-ux (parallel/before clarify)
    • epics.md - Created by /project-plan (comes AFTER clarify)

    Why: Clarify refines the INPUT (project.md). Don't confuse it with OUTPUT artifacts.

  3. Perform a structured ambiguity & coverage scan using this taxonomy. For each category, mark status: Clear / Partial / Missing. Produce an internal coverage map used for prioritization (do not output raw map unless no questions will be asked).

    Strategic Scope & Vision:

    • Project boundaries and success definition
    • Explicit exclusions and v2+ deferrals
    • Stakeholder alignment and priorities

    User & Market Understanding:

    • User segment definition and validation
    • Problem severity and frequency
    • Alternative solutions evaluation
    • Market timing and urgency

    Business Model & Constraints:

    • Revenue/cost model implications
    • Monetization strategy clarity
    • Pricing approach and validation plan
    • Unit economics assumptions (CAC, LTV)
    • Go-to-market strategy
    • Resource availability and skills
    • Timeline drivers and flexibility
    • Compliance and policy requirements

    Technical Landscape:

    • Integration requirements specificity
    • Performance and scale expectations
    • Security and privacy requirements
    • Platform and deployment constraints

    Risk & Dependencies:

    • Critical path dependencies
    • Technical feasibility concerns
    • Organizational change requirements
    • Market/competitive risks
  4. Generate clarifying questions:

    • Maximum 7 questions for project-level (strategic focus)
    • Prioritize by: Risk > Scope > Users > Technical > Other
    • Each question must be specific and actionable
    • Provide 2-3 answer options where helpful
    • Mark which section each answer will clarify

    Brownfield Adaptation:

    • If project-landscape-overview.md exists, SKIP questions about:
      • Existing features (already discovered in scan)
      • Current tech stack (already documented)
      • Current architecture patterns (already identified)
    • FOCUS brownfield questions on:
      • Strategic direction and future goals
      • User needs not evident in code
      • Business model and monetization
      • Constraints not visible in codebase (team, budget, timeline)
      • Planned enhancements or pivots
  5. Present questions professionally:

    code
    I've analyzed the project specification and identified some areas that would benefit from clarification before planning.
    
    1. **[Category] - [Specific Question]**
       Context: [Why this matters for project success]
       Options: 
       a) [Option 1]
       b) [Option 2]
       c) [Other/Custom]
       (This will clarify the [Section Name] section)
    
    [Continue for all questions...]
    
  6. Process each answer:

    • Acknowledge the response
    • Ask follow-up if answer is ambiguous
    • Identify which section(s) to update
  7. Update project.md systematically:

    • Add new "## Clarifications" section after Project Context if not exists
    • Create subsection "### Session YYYY-MM-DD"
    • List each Q&A pair concisely
    • Update relevant sections with clarified information
    • Remove any [NEEDS CLARIFICATION] markers that were resolved
    • Preserve all existing content structure
  8. Validate updates:

    • Re-scan for remaining ambiguities
    • Ensure internal consistency
    • Check no contradictions introduced
  9. Research need assessment (NEW - happens after Q&A):

    • Review remaining ambiguities after clarification
    • Identify which require research vs simple decisions
    • For each research need, determine:
      • Why it can't be answered via Q&A
      • What kind of research would help
      • Specific research queries to run
    • Categorize by priority: Critical, Important, Nice-to-have
  10. Update project.md with research section:

    • Add "## Research Needed" section if research identified
    • List each area with:
      • Topic and why it needs research
      • Priority level
      • Suggested research queries
    • This guides just-in-time research during downstream commands (ux/context/architecture/plan)
  11. Report completion:

  • Number of clarifications added
  • Sections updated
  • Research areas identified (if any)
  • Specific research queries prepared
  • Readiness assessment for next steps

Example interaction:

code
User: /project-clarify
AI: I've analyzed the project specification and identified some areas...

1. **Scope - Should the v1 include mobile apps or web-only?**
   Context: This significantly impacts timeline and resource needs.
   Options:
   a) Web-only for v1, mobile in v2
   b) Mobile-first, web later
   c) Both platforms from start
   (This will clarify the Project Type & Scale section)

User: Let's go with option a - web first, mobile later

AI: Understood. Web-only for v1 with mobile planned for v2.

[Continues through all questions...]

AI: ✅ Project Clarification Complete!

- Clarifications Added: 5
- Sections Updated: Project Type, Scope, Technical Constraints

Research Needed: 3 areas identified
1. **Market Analysis** (Critical)
   - Query: "AI code review tool competitive landscape 2024"
   - Why: Need competitive positioning before defining features

2. **Technical Feasibility** (Important)  
   - Query: "LLM fine-tuning costs and performance at scale"
   - Why: Architecture decisions depend on cost constraints

3. **User Validation** (Important)
   - Query: "Developer workflow integration patterns for AI tools"
   - Why: UX approach needs real-world usage patterns

Next Steps:
- If research needed: Apply the just-in-time research pattern
  (`.cursor/skills/just-in-time-research/SKILL.md`) to answer the queries above, then continue
- If UX-heavy product: /project-ux (define experience strategy)
- If standard product: /project-context (define constraints and standards)
- If very simple (Level 0-1): Skip to /project-plan

Note: Foundation commands (ux, context, constitution) provide essential
inputs for PRD creation. Running them before /project-plan ensures complete PRD.

Error conditions:

  • No project.md found → ERROR with instructions
  • Project already has PRD → WARN "Project already planned, clarifications may require re-planning"
  • User skips all questions → WARN about incomplete specification risks