brainstorming

Brainstorming Ideas Into Designs

Help turn ideas into fully formed designs and specs through natural collaborative dialogue.

Start by understanding the current project context, then ask questions one at a time to refine the idea. Once you understand what you're building, present the design and get user approval.

Do NOT invoke any implementation skill, write any code, scaffold any project, or take any implementation action until you have presented a design and the user has approved it. This applies to EVERY project regardless of perceived simplicity.

Anti-Pattern: "This Is Too Simple To Need A Design"

Every project goes through this process. A todo list, a single-function utility, a config change — all of them. "Simple" projects are where unexamined assumptions cause the most wasted work. The design can be short (a few sentences for truly simple projects), but you MUST present it and get approval.

Checklist

You MUST create a task for each of these items and complete them in order:

1. Choose artifact ID — create one shared YYYY-MM-DD-<topic-slug> ID for this whole workflow before creating any companion, spec, or plan files
2. Explore project context — check files, docs, recent commits
3. Create HTML visual companion (if topic will involve visual or highly structured review questions) — output the companion under docs/brainstorm/<artifact-id>/. See the Visual Companion and Structured HTML Companion sections below.
4. Ask clarifying questions — one at a time, understand purpose/constraints/success criteria
5. Propose 2-3 approaches — with trade-offs and your recommendation
6. Present design — in sections scaled to their complexity, get user approval after each section
7. Write design doc — save to docs/specs/<artifact-id>-design.md without committing
8. Spec review loop — dispatch spec-document-reviewer subagent with precisely crafted review context (never your session history); fix issues and re-dispatch until approved (max 3 iterations, then surface to human)
9. User reviews written spec and chooses escalation - ask user to review the spec file and, in the same prompt, decide whether to invoke grill-with-docs for a deeper pass against project language and documented decisions
10. Transition to implementation - invoke writing-plans skill to create implementation plan using the same artifact ID

Process Flow

digraph brainstorming {
    "Choose artifact ID" [shape=box];
    "Explore project context" [shape=box];
    "Visual questions ahead?" [shape=diamond];
    "Create HTML Visual Companion" [shape=box];
    "Ask clarifying questions" [shape=box];
    "Propose 2-3 approaches" [shape=box];
    "Present design sections" [shape=box];
    "User approves design?" [shape=diamond];
    "Write design doc" [shape=box];
    "Spec review loop" [shape=box];
    "Spec review passed?" [shape=diamond];
    "User reviews spec and wants grill-with-docs?" [shape=diamond];
    "Run grill-with-docs?" [shape=diamond];
    "Invoke grill-with-docs skill" [shape=box];
    "Invoke writing-plans skill" [shape=doublecircle];

    "Choose artifact ID" -> "Explore project context";
    "Explore project context" -> "Visual questions ahead?";
    "Visual questions ahead?" -> "Create HTML Visual Companion" [label="yes"];
    "Visual questions ahead?" -> "Ask clarifying questions" [label="no"];
    "Create HTML Visual Companion" -> "Ask clarifying questions";
    "Ask clarifying questions" -> "Propose 2-3 approaches";
    "Propose 2-3 approaches" -> "Present design sections";
    "Present design sections" -> "User approves design?";
    "User approves design?" -> "Present design sections" [label="no, revise"];
    "User approves design?" -> "Write design doc" [label="yes"];
    "Write design doc" -> "Spec review loop";
    "Spec review loop" -> "Spec review passed?";
    "Spec review passed?" -> "Spec review loop" [label="issues found,\nfix and re-dispatch"];
    "Spec review passed?" -> "User reviews spec and wants grill-with-docs?" [label="approved"];
    "User reviews spec and wants grill-with-docs?" -> "Write design doc" [label="changes requested"];
    "User reviews spec and wants grill-with-docs?" -> "Run grill-with-docs?" [label="approved"];
    "Run grill-with-docs?" -> "Invoke grill-with-docs skill" [label="yes"];
    "Invoke grill-with-docs skill" -> "Invoke writing-plans skill" [label="complete"];
    "Run grill-with-docs?" -> "Invoke writing-plans skill" [label="no"];
}

The terminal state is invoking writing-plans. Do NOT invoke frontend-design, mcp-builder, or any other implementation skill. The only optional skill between brainstorming and writing-plans is grill-with-docs, and only when the user explicitly chooses that escalation.

The Process

Understanding the idea:

• At the start of the workflow, choose one artifact ID: YYYY-MM-DD-<topic-slug> using the current date and a concise lowercase slug. Use the same ID for docs/brainstorm/<artifact-id>/, docs/specs/<artifact-id>-design.md, and the later implementation plan docs/plans/<artifact-id>.md.
• Check out the current project state next (files, docs, recent commits)
• Before asking detailed questions, assess scope: if the request describes multiple independent subsystems (e.g., "build a platform with chat, file storage, billing, and analytics"), flag this immediately. Don't spend questions refining details of a project that needs to be decomposed first.
• If the project is too large for a single spec, help the user decompose into sub-projects: what are the independent pieces, how do they relate, what order should they be built? Then brainstorm the first sub-project through the normal design flow. Each sub-project gets its own spec → plan → implementation cycle.
• For appropriately-scoped projects, ask questions one at a time to refine the idea
• Prefer multiple choice questions when possible, but open-ended is fine too
• Only one question per message - if a topic needs more exploration, break it into multiple questions
• Focus on understanding: purpose, constraints, success criteria

Tracking requirement sources:

As you ask questions and the user responds, mentally maintain two lists:

• User Requirements — things the user explicitly stated, chose from options you presented, or confirmed when you asked. These are facts.
• Agent Design Decisions — things you inferred, assumed, or recommended that the user did not explicitly request. These are your best guesses.

When proposing approaches: the user's choice of approach is a User Requirement. The implementation details you fill in around that choice are Agent Design Decisions.

When the user says "yes" or "looks good" to a design section: the items they originally requested remain User Requirements. The items you added remain Agent Design Decisions — approval of a section does not promote agent decisions to user requirements.

Exploring approaches:

• Propose 2-3 different approaches with trade-offs
• Present options conversationally with your recommendation and reasoning
• Lead with your recommended option and explain why

Presenting the design:

• Once you believe you understand what you're building, present the design
• Scale each section to its complexity: a few sentences if straightforward, up to 200-300 words if nuanced
• Ask after each section whether it looks right so far
• Cover: architecture, components, data flow, error handling, testing
• Be ready to go back and clarify if something doesn't make sense

Design for isolation and clarity:

• Break the system into smaller units that each have one clear purpose, communicate through well-defined interfaces, and can be understood and tested independently
• For each unit, you should be able to answer: what does it do, how do you use it, and what does it depend on?
• Can someone understand what a unit does without reading its internals? Can you change the internals without breaking consumers? If not, the boundaries need work.
• Smaller, well-bounded units are also easier for you to work with - you reason better about code you can hold in context at once, and your edits are more reliable when files are focused. When a file grows large, that's often a signal that it's doing too much.

Working in existing codebases:

• Explore the current structure before proposing changes. Follow existing patterns.
• Where existing code has problems that affect the work (e.g., a file that's grown too large, unclear boundaries, tangled responsibilities), include targeted improvements as part of the design - the way a good developer improves code they're working in.
• Don't propose unrelated refactoring. Stay focused on what serves the current goal.

After the Design

Documentation:

• Write the validated design (spec) to docs/specs/<artifact-id>-design.md
  • (User preferences for spec location override this default)
• Use elements-of-style:writing-clearly-and-concisely skill if available
• Do NOT commit the design document automatically; leave it uncommitted for user review unless the user explicitly asks you to commit

Spec document structure — requirement source separation:

The spec MUST have two clearly labeled top-level sections that separate what the user specified from what the agent filled in:

## User Requirements
Items the user explicitly stated, chose, or confirmed during brainstorming.
Each item should be traceable to something the user actually said or selected.

- [requirement]: [brief trace — e.g., "user's initial request", "chose from options", "confirmed when asked"]
- ...

## Agent Design Decisions
Everything the agent inferred, recommended, or filled in to complete the design.
Each item notes which user requirement it serves.

- [decision]: serves [which user requirement]. [rationale]
- ...

Rules for classification:

• If the user said it, chose it, or explicitly confirmed it as their own intent → User Requirement
• If the agent proposed it and the user said "looks good" or "yes" to a batch → Agent Design Decision (blanket approval does not promote agent decisions)
• For mixed items (user said "authentication," agent picked OAuth2): split them. "Authentication required" is a User Requirement. "OAuth2 with Google provider" is an Agent Design Decision that serves the authentication requirement.

Priority hierarchy: User Requirements are non-negotiable during implementation. Agent Design Decisions are flexible — they can be revised or dropped if they conflict with a User Requirement or if implementation reality demands it. This hierarchy carries forward into the plan and execution.

Spec Review Loop: After writing the spec document:

1. Dispatch spec-document-reviewer subagent (see spec-document-reviewer-prompt.md)
2. If Issues Found: fix, re-dispatch, repeat until Approved
3. If loop exceeds 3 iterations, surface to human for guidance

User Review Gate: After the spec review loop passes, ask the user to review the written spec and choose whether to run grill-with-docs before proceeding:

"Spec written to <path> and left uncommitted for your review. Please review it and let me know if you want to make any changes. Also, do you want to run grill-with-docs for a deeper pass against the project language, existing docs, and ADR-level decisions before we write the implementation plan?"

Wait for the user's response. If they request changes, make them and re-run the spec review loop. Only proceed once the user approves. If they approve the spec and say yes to grill-with-docs, invoke grill-with-docs and follow it before continuing. If they approve the spec and say no, proceed directly to writing-plans.

Implementation:

• Invoke the writing-plans skill to create a detailed implementation plan
• Do NOT invoke any other skill except the optional grill-with-docs escalation above. writing-plans is the next step after that choice is resolved.

Key Principles

• One question at a time - Don't overwhelm with multiple questions
• Multiple choice preferred - Easier to answer than open-ended when possible
• User requirements first - Never let agent design decisions override what the user explicitly said or chose. Track the source of every decision throughout the process.
• YAGNI ruthlessly - Remove unnecessary features from all designs
• Explore alternatives - Always propose 2-3 approaches before settling
• Incremental validation - Present design, get approval before moving on
• Be flexible - Go back and clarify when something doesn't make sense

Visual Companion

A browser-based companion for showing mockups, diagrams, visual options, and complex structured choices during brainstorming. Available as a tool — not a mode. Creating the companion means it's available for questions that benefit from visual or structured treatment; it does NOT mean every question goes through the browser.

Creating the companion: When you anticipate that upcoming questions will involve visual content (mockups, layouts, diagrams) or dense structured comparison, format your response in an HTML file. Do not ask whether the user wants a visual companion, and do not offer a text-only alternative.

Per-question decision: Decide FOR EACH QUESTION whether to use the browser or the terminal. The test: would the user understand this better by seeing it than reading it?

• Use the browser for content that IS visual or structurally dense — mockups, wireframes, layout comparisons, architecture diagrams, side-by-side visual designs, option comparison cards, decision matrices, architecture sketches, requirement grouping
• Use the terminal for content that is text — requirements questions, conceptual choices, short tradeoff lists, A/B/C/D text options, scope decisions

A question about a UI topic is not automatically a visual question. "What does personality mean in this context?" is a conceptual question — use the terminal. "Which wizard layout works better?" is a visual question — use the browser.

Before creating the HTML companion, read the detailed guide: skills/brainstorming/visual-companion.md

Structured HTML Companion

Use a Structured HTML Companion when the brainstorming output will be easier to review as a visual artifact than as prose. This is most useful for complex specs, many options, or decisions with multiple tradeoffs.

HTML companion is a review aid, not the canonical spec. The Markdown spec in docs/specs/<artifact-id>-design.md remains the source of truth unless the user explicitly asks otherwise.

Good structured companion sections:

• User Requirements — grouped by theme or priority
• Agent Design Decisions — linked to the user requirement each decision serves
• Open Questions — unresolved choices needing user input
• Approach Comparison — recommended option plus alternatives
• Risks / Tradeoffs — visible costs, constraints, and mitigations

Create a structured companion for:

• option comparison cards
• decision matrices
• architecture sketches
• requirement grouping
• multi-step user journeys
• large design reviews where scanning matters

Skip HTML for short answers, ordinary clarification questions, simple scope decisions, commands, logs, or anything that is clearer as plain Markdown.