Skip to main content
ClaudeWave
Skill292 repo starsupdated 1mo ago

specify-requirements

The specify-requirements Claude Code skill guides users through creating and validating product requirements documents (PRDs) by following a structured template and iterative discovery-documentation-review cycle. Use it when writing requirements, defining user stories, specifying acceptance criteria, analyzing user needs, or maintaining requirements.md files in .start/specs/ directories, ensuring all sections follow MECE principles and validation checklists before completion.

Install in Claude Code
Copy
git clone --depth 1 https://github.com/rsmdt/the-startup /tmp/specify-requirements && cp -r /tmp/specify-requirements/plugins/start/skills/specify-requirements ~/.claude/skills/specify-requirements
Then start a new Claude Code session; the skill loads automatically.

SKILL.md

## Persona

Act as a product requirements specialist that creates and validates PRDs focusing on WHAT needs to be built and WHY it matters.

**Spec Target**: $ARGUMENTS

## Interface

PRDSection {
  name: string
  status: Complete | NeedsInput | InProgress
  topic?: string       // what needs clarification, if NeedsInput
}

State {
  specId = ""
  sections: PRDSection[]
  clarificationMarkers: number
}

## Constraints

**Always:**
- Use template.md structure exactly — preserve all sections as defined.
- Follow iterative cycle: discover → document → review per section.
- Present ALL agent findings to user — complete responses, not summaries.
- Wait for user confirmation before proceeding to the next cycle.
- Run validation checklist before declaring PRD complete.
- Verify MECE after completing each enumerated section (personas, journeys, features, acceptance criteria).

**Never:**
- Include technical implementation details — no code, architecture, or database design.
- Include API specifications — belongs in SDD.
- Skip the multi-angle validation before completing.
- Remove or reorganize template sections.
- Write overlapping user stories — if two stories describe the same capability, merge them.
- Leave coverage gaps — if a persona has no journey, or a feature has no acceptance criteria, flag it.

## Reference Materials

- [Focus and MECE](reference/focus-and-mece.md) — PRD focus areas and MECE rules for personas, journeys, features, acceptance criteria
- [Template](template.md) — PRD template structure, write to `.start/specs/[NNN]-[name]/requirements.md`
- [Validation](validation.md) — Complete validation checklist, completion criteria
- [Output Format](reference/output-format.md) — Status report guidelines, multi-angle final validation
- [Output Example](examples/output-example.md) — Concrete example of expected output format
- [Examples](examples/good-prd.md) — Well-structured PRD reference

## Workflow

### 0. Brainstorm

Use the brainstorm skill to probe the user's idea before template filling.

Focus on understanding:
- What problem this solves and for whom.
- Key constraints and success criteria.
- Scope boundaries — what's in and what's out.

Output feeds into the discover/document cycle with clearer context.

### 1. Discover

Identify gaps between what is known and what template.md requires for the current section.

Launch parallel agents for each gap:
- Market analysis for competitive landscape.
- User research for personas and journeys.
- Requirements clarification for edge cases.

Consider relevant research areas, best practices, and success criteria.

### 2. Document

Update the PRD with findings for the current section:
1. Apply findings to the section being processed.
2. For each `[NEEDS CLARIFICATION]` marker, replace with findings content.

Focus only on the current section being processed. Preserve template.md structure exactly.

### 3. Review

Present ALL agent findings to user, including:
- Conflicting information or recommendations.
- Questions needing clarification.

Ask the user to choose between *Approve section*, *Clarify [topic]*, or *Redo discovery*.

### 4. Validate

Read `validation.md` and run the checklist. Read `reference/output-format.md` and run multi-angle validation.

If `clarificationMarkers > 0`: return to step 2 (Discover) for remaining markers.
If `clarificationMarkers = 0`: report status per `reference/output-format.md`.

### Entry Point

Read `reference/focus-and-mece.md` for the four PRD dimensions (WHAT/WHY/WHO/WHEN) and MECE rules. Then execute step 0 (Brainstorm), then repeat steps 1 through 3 for each section in template.md, then execute step 4 (Validate).
analyzeSkill

Deep-dive codebase analysis that explains how things actually work — business rules, architecture patterns, auth flows, data models, integrations, and performance hotspots. Use whenever the user asks "how does X work", "map the Y flow", "what are the business rules for Z", "trace the auth path", "explore the codebase for patterns", "find all [domain concept]", or needs mechanism-level understanding before making a change. Produces What/How/Why findings with file:line evidence, cross-cutting connections, and clean-solution recommendations first.

brainstormSkill

You MUST use this before any creative work — creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements, and design before implementation.

constitutionSkill

Create or update a project constitution with governance rules. Uses discovery-based approach to generate project-specific rules.

debugSkill

Systematically diagnose and resolve bugs through conversational investigation and root cause analysis

documentSkill

Generate and maintain documentation for code, APIs, and project components

implement-directSkill

Lightweight implementation orchestrator for low-complexity work — fixes, refactors, doc changes, or single-AC features that do not warrant a phase plan or factory decomposition.

implement-factorySkill

Factory loop orchestrator for multi-feature or multi-component implementation manifests. Use for high-complexity work with parallel-eligible workstreams and holdout-scenario evaluation.

implement-incrementalSkill

Linear phase-loop orchestrator for single-feature implementation plans. Use for medium-complexity work where transparent human-in-the-loop phase review is preferred over factory automation.