"Clarify ambiguous work through structured brainstorming: ask focused questions, propose 2-3 approaches with tradeoffs, and produce a concise design brief (goals, non-goals, constraints, risks, validation). Use when requirements are unclear or before large/irreversible changes."
Resources
1Install
npx skillscat add dmonteroh/curated-agent-skills/brainstorming Install via the SkillsCat registry.
SKILL.md
Brainstorming: Ideas → Design Brief
Use this skill when
- The user’s requirements are ambiguous, conflicting, or incomplete
- The cost of “building the wrong thing” is high (irreversible changes, migrations, public APIs)
- A fast, structured way to compare 2–3 approaches and pick one is needed
Do not use this skill when
- The request is already clear enough to plan and implement directly
- The user explicitly requests immediate implementation without upfront design
- The work is tiny/low-risk and a design brief would be overhead
Required inputs
- The user's current goal or problem statement (even if vague)
- Known constraints (time, scope, tech, risk, compliance)
- Success criteria or definition of done (if known)
- Stakeholders/users and any explicit non-goals (if known)
Overview
Provides a structured dialogue that turns fuzzy ideas into a clear, testable design brief.
Default posture is to clarify and avoid premature implementation.
Workflow
1) Understand the idea (question loop)
- Briefly inspects current project context (relevant docs/files) when available.
- Asks one question at a time; prefers multiple choice when it speeds decisions.
- Clarifies: purpose, users, constraints, success criteria, non-goals.
- Output: a short problem statement and list of open questions.
2) Explore approaches
- Proposes 2-3 approaches with tradeoffs against the stated constraints.
- Makes a recommendation and explains why it wins for this context.
- Output: 2–3 options with pros/cons/risks and a recommendation.
3) Present the design brief (incremental validation)
- Presents a short design brief and asks for confirmation.
- If the user disagrees, returns to the question loop and iterates.
- Covers only what matters for decision-making:
- goals / non-goals
- scope boundaries
- main flows and key states
- risks + mitigations
- validation (how we know it worked)
- Output: a design brief plus a confirmation request.
4) Decide next step
- If the user is ready, asks to move to execution planning.
- If not, continues the question loop.
- Output: a single next-step question.
Decision points
- If constraints make all options invalid, ask which constraint can change.
- If the user rejects the recommendation, return to step 2 with updated criteria.
- If the user wants to implement, move to execution planning.
After the Design
Documentation (optional)
- If the repo has a preferred planning/spec protocol, adapts the brief to that format.
- Otherwise, writes a design note to an agreed path (example):
docs/plans/YYYY-MM-DD-<topic>-design.mdDoes not assume other skills exist; treats any integrations as optional.
Implementation (optional)
- Asks: "Ready to move from design to execution?"
- If proceeding, produces a small execution plan (milestones + verification).
Key Principles
- One question at a time - Avoids overwhelming with multiple questions
- Multiple choice preferred - Keeps answers easier than open-ended when possible
- YAGNI ruthlessly - Removes unnecessary features from all designs
- Explore alternatives - Proposes 2-3 approaches before settling
- Incremental validation - Presents design in sections and validates each
- Be flexible - Returns to clarification when something doesn't make sense
Common pitfalls
- Skipping constraints and non-goals during questioning
- Listing options without explaining tradeoffs
- Producing a design brief without asking for confirmation
- Moving into implementation without explicit user approval
Examples
Input
"We need to improve onboarding but aren't sure what to build yet. Can you help us pick a direction?"
Output (abridged)
- Problem & success: reduce drop-off in first session by 20% without increasing support load.
- Approaches:
- Guided walkthrough (pros: fast; cons: brittle)
- Goal-based checklist (pros: flexible; cons: requires UX work)
- Recommendation: checklist, aligns with user diversity.
- Design brief: goals, non-goals, constraints, risks, verification.
- Next step: "Want an execution plan next?"
Output Contract (Always)
- A 3–7 bullet problem statement + success criteria
- 2–3 approaches with pros/cons and a recommendation
- A design brief with explicit non-goals, risks, and a verification plan
- A single next-step question
Reporting format
## Problem & Success
- ...
## Approaches
1) Option A — pros/cons/risks
2) Option B — pros/cons/risks
Recommendation: ...
## Design Brief
- Goals:
- Non-goals:
- Constraints:
- Key flows/states:
- Risks & mitigations:
- Verification plan:
## Next Step
- ...References (Optional)
references/implementation-playbook.md
Scripts
- None. Use instructions only unless the user requests automation.