clarify
Review the current feature specification for product clarity, completeness, and edge cases. Product-focused only — no technical or implementation questions.
Best use case
clarify is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Review the current feature specification for product clarity, completeness, and edge cases. Product-focused only — no technical or implementation questions.
Teams using clarify should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/clarify/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How clarify Compares
| Feature / Agent | clarify | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Review the current feature specification for product clarity, completeness, and edge cases. Product-focused only — no technical or implementation questions.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Clarify Skill
Review specifications for **product** clarity and completeness. Focus exclusively on what the user experiences — not how it's built.
## When to Activate
This skill is relevant when:
- Reviewing product specifications for ambiguity or gaps
- Checking that user-facing behavior is fully defined
- Validating feature scope and product value
- Identifying missing user flows and edge cases
## Output File
Always write the review to a markdown file alongside the spec:
- Path: `<spec-directory>/clarification-review.md`
- Every gap, question, and missing scenario must include a **Recommendation** — a concrete suggestion for what to add or change in the spec to resolve it.
- End the file with a **Verdict**: `APPROVED` (no critical gaps) or `NEEDS REVISION` (critical/important gaps found).
## Scope Boundary
**IN SCOPE — Product questions:**
- User flows, interactions, and behavior
- What the user sees, clicks, and experiences
- Business logic and rules
- Edge cases from the user's perspective
- Success metrics and acceptance criteria
- UX copy, states (empty, error, loading, success)
- Feature scope and prioritization
**OUT OF SCOPE — Do NOT raise these:**
- API design, endpoint structure, or request/response contracts
- Database schema, data models, or type definitions
- Code architecture, file structure, or refactoring concerns
- Performance implementation details (model choice, caching strategy)
- Prompt engineering or prompt structure
- Library/framework choices
- Existing code gaps or tech debt
- Anything that belongs in a technical plan, not a product spec
When in doubt, ask: "Would a product manager care about this?" If the answer is no, skip it.
## Core Principles
### Value Critical
- Does this solve a real user problem?
- Is the "why" clear?
- Strategic alignment with product vision
- Worth building?
### Clarity & Completeness
- No ambiguous terms like "fast" or "easy"
- Detailed acceptance criteria for user-facing behavior
- Specific, measurable requirements
- All user scenarios covered
### Scope Check
- Is this MVP or full roadmap?
- Should we cut scope to ship faster?
- Are there product blockers or dependencies?
- Realistic given constraints
### Strategic Alignment
- Fits broader product vision
- Right priority for now
- Complements existing features
## Quick Checks
When reviewing specifications, verify:
- [ ] Feature descriptions are unambiguous
- [ ] User-facing requirements are actionable
- [ ] Empty states defined
- [ ] Error states defined (what the user sees)
- [ ] Loading states defined (what the user sees)
- [ ] Success states defined
- [ ] No "happy path" bias
- [ ] Vague requirements challenged ("make it pop", "good UX")
- [ ] Cross-referenced with existing product capabilities
- [ ] MVP scope is realistic
- [ ] Missing user flows identified
- [ ] Edge cases documented from the user's perspective
- [ ] Acceptance criteria are testable and user-observable
- [ ] Every gap has a concrete Recommendation
- [ ] Review file written to `<spec-directory>/clarification-review.md`
- [ ] Verdict included at end of review file