ワンクリックで
brainstorming
Turn ideas into fully formed designs through collaborative dialogue
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Turn ideas into fully formed designs through collaborative dialogue
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when working with Jujutsu (jj) version control system, creating changes, managing bookmarks, or pushing code. Triggers on jj commands and .jj/ directory presence.
Provides Vacanti's flow metrics principles when discussing predictability, forecasting, cycle time, throughput, or delivery estimates. Triggers on keywords like "estimate", "when will", "how long", "velocity", "average cycle time", "story points", "deadline", "SLE", "percentile", "forecast"
Guides appropriate responses to process variation using Vacanti's framework. Triggers when discussing process improvement, responding to metric changes, XmR chart signals, sprint retrospectives, or when someone asks "why did our velocity/throughput change?"
This skill should be used when performing browser-based exploratory testing of web applications, including functional testing, visual inspection, accessibility checks, and performance analysis. Triggers when testing web UIs, checking for browser bugs, validating WCAG compliance, or measuring Core Web Vitals.
This skill should be used when creating plan files, writing Gherkin requirements, or parsing plan file structure. Provides the plan file template and format specifications.
Opinionated REST API design guidance for resource naming, HTTP semantics, error formats, pagination, versioning, and more. Use when designing REST APIs during brainstorming or plan creation.
| name | brainstorming |
| description | Turn ideas into fully formed designs through collaborative dialogue |
Help turn ideas into fully formed designs and specs through natural collaborative dialogue.
Check project context first:
Ask questions one at a time:
Key questions to explore:
Propose 2-3 different approaches:
Present options conversationally:
I see three ways to approach this:
**Option 1: [Name]** (Recommended)
[Description]
- Pro: [advantage]
- Con: [disadvantage]
**Option 2: [Name]**
[Description]
- Pro: [advantage]
- Con: [disadvantage]
**Option 3: [Name]**
[Description]
- Pro: [advantage]
- Con: [disadvantage]
I recommend Option 1 because [reasoning].
Once you understand what to build:
Section-level expert review:
Before the user sees each design section, route it to relevant experts based on content keywords:
| Section Content Keywords | Expert(s) Invoked |
|---|---|
| database, schema, migration, storage, persistence, query | Data Systems Expert |
| page, form, button, display, UI, component, dashboard, click | UX Expert + Accessibility Expert |
| endpoint, request, response, REST, API, HTTP | REST API Guidelines skill + Security Expert |
| deploy, pipeline, build, release, CI/CD, container | Continuous Delivery Expert + Cloud Expert |
| test, spec, scenario, assert, mock, stub, fixture | Testing Expert |
| module, service, boundary, integration, composition, coupling | Software Architecture Expert |
| user, PII, data collection, consent, retention, GDPR | Privacy Expert |
| metric, SLO, monitor, alert, log, trace, observe | SRE Expert |
| domain, entity, aggregate, bounded context, ubiquitous language | Domain Design Expert |
| lead time, deployment frequency, feedback loop, batch size | Engineering Effectiveness Expert |
| (always) | Product Strategy Expert — reviews every section for product value |
For each section:
> Reviewed by Data Systems Expert: Suggested adding index strategy for the query pattern described.
> Reviewed by Product Strategy Expert: Confirmed this delivers clear user value.
This adds latency to brainstorming but produces higher-quality designs. The end-of-plan Gherkin review (/review-features) still runs as a final pass, but the heavy lifting happens section-by-section here.
Sections to cover:
When invoked from the plan command (/tommymorgan:plan):
When invoked standalone:
<project>/plans/YYYY-MM-DD-<topic>.mdImplementation (if continuing):
/tommymorgan:work to execute the planOne question at a time Don't overwhelm with multiple questions. If a topic needs exploration, break it into sequential questions.
Multiple choice preferred Easier to answer than open-ended. Use when you can enumerate reasonable options.
YAGNI ruthlessly Remove unnecessary features from all designs. Build what's needed now, not what might be needed later.
Explore alternatives Always propose 2-3 approaches before settling. The first idea is rarely the best.
Incremental validation Present design in sections. Validate each before moving on. Catch misunderstandings early.
Be flexible Go back and clarify when something doesn't make sense. The goal is the right design, not finishing fast.