brainstorming
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.
- License 许可
- Apache-2.0
- 所属仓库
- zak-keown/moe
- 最近更新
- 2026-09-05
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- brainstorming
- triggers
- Load when starting new creative work that needs design exploration before implementation — new features, components, or behavioral changes. Do NOT load for: bug fixes, refactoring existing code, test additions, documentation edits, or tasks with an approved spec already in hand (use `write-plan` directly).
- description
- 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.
02 资源文件 7
- 脚本scripts/frame-template.html7.9 KB
- 脚本scripts/helper.mjs4.9 KB
- 脚本scripts/server.mjs20.5 KB
- 脚本scripts/start-server.mjs7.1 KB
- 脚本scripts/stop-server.mjs3.8 KB
- 参考文档spec-document-reviewer-prompt.md1.7 KB
- 参考文档visual-companion.md13.3 KB
03 SKILL.md 正文
Brainstorming Ideas Into Designs
Help turn ideas into fully formed designs and specs through natural collaborative dialogue.
Start by classifying the request by DEPTH, then work through the depth's process: understand the context, refine the idea, present a design, and get your human partner's approval.
<HARD-GATE> Do NOT invoke any implementation skill, write any code, scaffold any project, or take any implementation action until you have told your human partner what you intend and they have approved it. This applies to EVERY task at EVERY depth below — the ceremony scales with the task; the approval gate never does. </HARD-GATE>
Stress-testing an existing design
Before classifying by depth, check: is the user arriving with a substantially
complete design, plan, or decision and asking to validate it — "grill me on
this," "stress-test this plan," "poke holes in this," "what am I missing"? If
so, skip depth classification entirely. Read the design, build the design tree
from it, and enter the Sharpening an agreed shape round loop directly. The
<HARD-GATE> still holds: when rounds are done, present what held up and what
didn't, then get approval before any implementation.
If a gap discovered during stress-testing upgrades the scope — the design needs a new subsystem, or an assumption collapses — stop the round loop, re-enter depth classification at the appropriate depth, and continue from there.
Three Depths
Before your first question, classify the request by DEPTH and say the
classification out loud — "this looks like a change, so I'll present
a short design here rather than write a spec" — so your human partner
can override it. The three depths, in increasing size, are patch,
change, and feature.
- Patch — the smallest depth: a one-line fix, a config tweak, a targeted spot-tightening, or a quick feasibility probe whose output is an answer, not code you keep. Present the intent in 1-2 sentences, get a nod, then keep it minimal. No design doc, no spec file. If a probe yields code you now want to keep, that is a NEW request — reclassify it and get its own approval.
- Change — a well-scoped modification to code that already exists
in this repo: a new flag, a small endpoint, a small feature spanning
a few files. Familiarity with the KIND of app is not enough — the
flow you are changing must already be here to read. If there is no
existing flow to change, the depth is not
change. Ask the clarifying questions that matter, present a short design IN CHAT (a few sentences to a few short paragraphs), and STOP. Implementation starts only after your human partner says yes to that design — achange-depth approval is as hard a gate as afeature-depth one. No spec file, no implementation plan document. - Feature — new projects, new subsystems, changes that restructure how components fit together or alter interfaces others depend on. Follow the full process: questions, approaches, sectioned design, written spec, then the write-plan skill.
When in doubt between two depths, take the heavier one. The ratchet is one-way: hidden complexity discovered mid-task upgrades the depth — stop, say so, and step up. Nothing downgrades mid-task.
Anti-Pattern: "Too Simple To Need Approval"
Every depth ends with your human partner approving your intent before implementation. A todo list, a single-function utility, a config change — the design may be two sentences in chat, but you MUST present it and get approval. "Simple" tasks are where unexamined assumptions cause the most wasted work. What scales with simplicity is the artifact, never the approval.
Red Flags
| Thought | Reality |
|---|---|
| "This is too simple to need a design" | Simple means a short design, not no design. Two sentences in chat, then approval. |
"I'll call it a change and skip the spec" |
Reaching for a label to skip work IS the doubt — take the heavier depth. |
"It's a change and the design is obvious — I'll start while they read it" |
The gate is the approval, not the design's length. Present, then stop until you hear yes. |
"I understand this kind of app, so it's a change" |
change measures the repo, not your familiarity. A new project has no existing flow — it is a feature. |
"The patch probe worked, so I'll keep the code" |
A patch probe's deliverable is an answer, not code. Keeping the code is a new request — reclassify. |
| "It grew, but I'm almost done — no need to reclassify" | Hidden complexity upgrades the depth mid-task. Stop and say so. |
"They approved the patch, so the follow-up work is approved too" |
Each task gets its own depth and its own approval. |
| "The shape is agreed, so I can just start" | An agreed shape is not a set of resolved decisions. Round the frontier, present, then get the yes. |
| "One question per message, so these six take six turns" | If they do not depend on each other, that is six round-trips doing one round's work. Ask the frontier together. |
"This patch deserves a helper class and a full test suite" (gold-plating) |
A patch is the minimum change that solves the task, plus the test for what actually changed. Extra structure is scope creep at the smallest depth. |
"The feature's interfaces exist; I'll wire them up later" (stub-and-declare) |
A feature is complete when the flow runs end-to-end, not when the skeleton compiles. Sketching structure and declaring done is the feature-depth failure mode. |
Checklist
Classify first, announce the depth, then use TodoList to manage each item
on that depth's list and complete them in order.
Patch:
- Explore project context — enough to frame the change or probe
- Present intent + minimal plan — 1-2 sentences
- Get approval — a nod is enough
- Do the smallest thing that solves it — as cheaply as correctness allows; a probe's deliverable is an answer, a fix's deliverable is the fix plus the test for what actually changed
- Report — the outcome, or the recommendation if the deliverable was an answer; label any exploratory code as throwaway
Change:
- Explore project context — check files, docs, recent commits
- Ask clarifying questions — one at a time, the ones that matter
- Present short design in chat — approach, files touched, testing
- Get approval — STOP and wait for an explicit yes; presenting the design and starting in the same breath is skipping the gate
- Implement — proceed with the normal development workflow (TDD applies); no plan document
Feature:
- Explore project context — check files, docs, recent commits
- Open the visual companion just-in-time — NOT upfront. The first time a question would genuinely be clearer shown than described, generate the visual, start the companion with
--open, and present it in the browser. If no visual question ever arises, never start it. See the Visual Companion section below. - Ask clarifying questions — one at a time, understand purpose/constraints/success criteria
- Propose 2-3 approaches — with trade-offs and your recommendation
- Present design — in sections scaled to their complexity, get user approval after each section
- Write design doc — use
moe jig spec init <topic>to createdocs/moe/specs/YYYY-MM-DD-<topic>-design.md, then fill in the content and commit - Spec self-review — quick inline check for placeholders, contradictions, ambiguity, scope (see below)
- User reviews written spec — ask user to review the spec file before proceeding
- Transition to implementation — invoke write-plan skill to create implementation plan
Process Flow
digraph brainstorming {
"Classify: patch / change / feature" [shape=diamond];
"Present intent + minimal plan (1-2 sentences)" [shape=box];
"Ask clarifying questions (change)" [shape=box];
"Present short design in chat" [shape=box];
"Human approves?" [shape=diamond];
"Do smallest thing; report" [shape=doublecircle];
"Implement via normal workflow (no plan doc)" [shape=doublecircle];
"Explore project context" [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 self-review\n(fix inline)" [shape=box];
"User reviews spec?" [shape=diamond];
"Invoke write-plan skill" [shape=doublecircle];
"Hidden complexity? Upgrade depth" [shape=box];
"Classify: patch / change / feature" -> "Present intent + minimal plan (1-2 sentences)" [label="patch"];
"Classify: patch / change / feature" -> "Ask clarifying questions (change)" [label="change"];
"Classify: patch / change / feature" -> "Explore project context" [label="feature"];
"Present intent + minimal plan (1-2 sentences)" -> "Human approves?";
"Ask clarifying questions (change)" -> "Present short design in chat";
"Present short design in chat" -> "Human approves?";
"Human approves?" -> "Do smallest thing; report" [label="patch: yes"];
"Human approves?" -> "Implement via normal workflow (no plan doc)" [label="change: yes"];
"Hidden complexity? Upgrade depth" -> "Classify: patch / change / feature";
"Explore project context" -> "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 self-review\n(fix inline)";
"Spec self-review\n(fix inline)" -> "User reviews spec?";
"User reviews spec?" -> "Write design doc" [label="changes requested"];
"User reviews spec?" -> "Invoke write-plan skill" [label="approved"];
}
The graph routes by depth only. Sharpening an agreed shape is orthogonal to it — a way of running the question steps, not another branch out of the classifier — so it does not appear as a node.
Terminal states are depth-bound. feature: the ONLY skill you
invoke after brainstorming is write-plan — never frontend-design,
mcp-builder, or any other implementation skill. change: after
approval, implementation proceeds directly through the normal
development workflow; no plan document. patch: the terminal state is
the smallest change that solves the task, or — for a patch-shaped
probe — a reported recommendation.
The Process
The subsections below serve the change and feature depths (a
patch stops at "present the intent, get a nod, do the smallest
thing"). Sections from Exploring approaches onward are
feature-depth material — for change work, context plus a few
questions plus a short in-chat design is the whole process.
Understanding the idea:
- Check out the current project state first (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 via ask the user in chat when possible, but open-ended is fine too
- Only one question per message while the shape is still moving - if a topic needs more exploration, break it into multiple questions. Once the shape is agreed and only decisions remain, switch to rounds (see Sharpening an agreed shape below)
- Focus on understanding: purpose, constraints, success criteria
Sharpening an agreed shape:
Sometimes the shape is not what is in question. Your human partner arrives with the design already settled, or you have just been given approval on it, and what is left is a set of decisions to nail down. Asking those one at a time is the wrong instrument: it serialises questions that do not depend on each other, and burns a round-trip on each.
Switch to rounds. Map what remains as a design tree — every decision branches into the decisions that hang off it. The frontier is every decision whose prerequisites are already settled: the questions you can ask now without guessing at answers you have not heard yet.
Ask the whole frontier in one round. Number each question and give your recommended answer:
Q1 — <short title>: <the question, with the options if there are any>
Recommendation: <your answer, and why>
Q2 — <short title>: …
Recommendation: …
Then stop and wait. Each round of answers reshapes the tree: settled decisions push the frontier outward and unblock the questions that depended on them. Recompute the frontier and ask the next round. A question whose answer depends on another question still open in this round belongs to a later round, not this one.
Facts are yours to find; decisions are your human partner's. When a frontier question turns on a fact from the environment — what is in the file, what the tool returns — go and get it. Never ask for something you could look up. Do not block the round on it either: an exploration still running is an unsettled prerequisite, so only the questions downstream of it wait. Ask the rest now.
The round loop is done when the frontier is empty — every branch of the tree
visited, nothing left silently assumed. Then present the design as usual. This
mode changes how you ask, never whether you get approval: the <HARD-GATE>
holds exactly as it does everywhere else.
Applies at change and feature. A patch has no frontier worth mapping —
present the intent, get a nod, do the smallest thing.
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
- YAGNI ruthlessly - remove unnecessary features from every approach and design
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 (feature depth)
Documentation:
- Create the spec file with
moe jig spec init <topic>— it writesdocs/moe/specs/YYYY-MM-DD-<topic>-design.mdwith the correct naming and a skeleton. Fill in the validated design and commit.- (User preferences for spec location override this default)
- Use the
write-clearlyskill (a sibling skill in this plugin) - Commit the design document to git
Spec Self-Review: After writing the spec document, look at it with fresh eyes:
- Placeholder scan: Any "TBD", "TODO", incomplete sections, or vague requirements? Fix them.
- Internal consistency: Do any sections contradict each other? Does the architecture match the feature descriptions?
- Scope check: Is this focused enough for a single implementation plan, or does it need decomposition?
- Ambiguity check: Could any requirement be interpreted two different ways? If so, pick one and make it explicit.
Fix any issues inline. No need to re-review — just fix and move on.
User Review Gate: After the spec review loop passes, ask the user to review the written spec before proceeding:
"Spec written and committed to
<path>. Please review it and let me know if you want to make any changes before we start writing out 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.
Implementation:
- Invoke the write-plan skill to create a detailed implementation plan
- Do NOT invoke any other skill. write-plan is the next step.
Visual Companion
A browser-based companion for showing mockups, diagrams, and visual options during brainstorming. Available as a tool — not a mode. Starting the companion does NOT mean every question goes through the browser.
The browser companion is rung 2 of the shared native-rendering ladder in skills/_shared/native-rendering.md, resolved relative to this loaded document.
Rung 1 (the Claude Code Artifact tool) is not exposed — start at
rung 2 (the brainstorm browser companion) and drop to rung 3 (local
HTML file) or rung 4 (markdown file) when a browser is unavailable.
Opening the companion (just-in-time): Do NOT start it upfront. Wait until a question would genuinely be clearer shown than told — a real mockup / layout / diagram question, not merely a UI topic. The first time that happens, generate the visual, start the server with --open, and present the question there immediately. Do not warn about the companion or ask for separate approval before opening it.
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 — mockups, wireframes, layout comparisons, architecture diagrams, side-by-side visual designs
- Use the terminal for content that is text — requirements questions, conceptual choices, 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 first use, read the detailed guide:
skills/brainstorming/visual-companion.md
04 同仓库更多 Skills
write-skill
Use when creating new skills, editing existing skills, or verifying skills work before deployment
write-plan
Use when you have a spec or requirements for a multi-step task, before touching code
write-clearly
Apply Strunk's timeless writing rules to ANY prose humans will read—documentation, commit messages, error messages, explanations, reports, or UI text. Makes your writing clearer, stronger, and more professional.
windows-vm
Create, manage, or connect to a headless Windows 11 VM running in Docker with SSH access. Use when the user wants to spin up, stop, restart, or SSH into a Windows VM.
verify-completion
Use when about to claim work is complete, fixed, or passing, before committing or creating MRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
using-moe
Use when starting any conversation - establishes how to find and use skills, requiring skill invocation before ANY response including clarifying questions