hardness1020 clarify-demand
Conversational demand clarification and feasibility research before formal work item registration.
Purpose
This skill helps users go from a vague idea to a concrete work item registration. It runs before /manage-work register to ensure the right track, scope, and description are chosen.
Works on any branch (including main).
4-Phase Workflow
Phase 1: Demand Clarification
Ask the user to clarify the following (skip items they've already provided):
- Problem — What problem are you solving? What's broken or missing?
- Users — Who benefits from this? (end users, developers, ops, etc.)
- Done Criteria — How will you know it's done? What does success look like?
- Constraints — Any technical, time, or compatibility constraints?
- Scope Signal — Is this a bug fix, a single feature, a multi-component change, or a system-level change?
Keep the conversation natural. Ask follow-up questions if answers are vague. One or two rounds is usually enough.
Phase 2: Feasibility Research
Perform read-only codebase analysis:
- Does existing code support this, or is new infrastructure needed?
- What files/modules would be affected?
- Are there existing patterns to follow?
- What are the risks (high coupling, missing tests, architectural debt)?
Skip this phase for obvious Micro/Small items where the scope is clear and the change is localized.
Phase 3: Synthesis
Produce an Intake Brief (displayed in conversation, not persisted to disk):
## Intake Brief
**Problem**: [1-2 sentences]
**Users**: [who benefits]
**Feasibility**: [summary from Phase 2, or "Skipped — scope is clear"]
**Complexity**: [Low / Medium / High]
**Recommended Track**: [Micro / Small / Medium / Large]
**Rationale**: [why this track]
**Suggested Description**: "<description for register command>"
**Suggested ID**: <numeric ID>
Phase 4: Handoff
Present the Intake Brief to the user. If they confirm:
/manage-work register "<description>" <ID> <track>
If they want changes, loop back to the relevant phase.
Guidelines
- Ephemeral output: The Intake Brief lives in the conversation only — no files are written.
- Bias toward smaller tracks: When in doubt, recommend Small over Medium. Users can always upgrade.
- Don't over-research: Phase 2 should take 1-2 minutes, not 10. Quick scan, not deep dive.
- Natural language: This is a conversation, not a form. Adapt to the user's communication style.