Token导航 LogoToken导航

Agent Skill · 研究检索

brainstorming(头脑风暴)

brainstorming 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

brainstorming 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 48,042。

总安装 706周安装 30GitHub Stars 1下载 247核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:brainstorming(头脑风暴)
来源仓库:https://github.com/pixel-process-ug/superkit-agents
仓库路径:skills/brainstorming
安装命令:
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill brainstorming
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill brainstorming

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

brainstorming 用于查找、检索和筛选相关信息,适合快速定位候选结果。适用于需要根据关键词或任务场景从来源线索中筛选信息的场景。通过 npx skills add 命令从指定 GitHub 仓库安装并使用该技能。安装前需确认权限范围和维护状态,注意可能触发联网或文件读写操作。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/brainstorming

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill brain…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

brainstorming 怎么安装?

在终端执行 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill brainstorming。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

brainstorming 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

brainstorming 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

brainstorming 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

brainstorming 用的人多吗?

公开数据是累计安装 706 次,近一周安装 30 次,下载 247 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

brainstorming 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
brainstorming

头脑风暴 · 本页

7061只读未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Brainstorming Ideas Into Designs

Overview

Brainstorming transforms vague ideas into validated designs through structured collaborative dialogue. It bridges the gap between "I want X" and a concrete design that can be handed to the planning skill. This enhanced version integrates with the self-learning skill to build on known project context and avoid re-asking previously answered questions.

Announce at start: "I'm using the brainstorming skill to explore this idea and create a design before implementation."

Trigger Conditions

  • User describes a new feature idea or goal
  • User asks "how should we build X?"
  • Creative work requires design exploration before planning
  • /brainstorm command invoked
  • Transition from a vague requirement to structured design is needed

Phase 1: Context Loading

Goal: Load everything known about the project before asking the user anything.

  1. Check memory files (project-context.md, learned-patterns.md, decisions-log.md) for known patterns, stack, and conventions
  2. Read relevant existing code, docs, and recent commits
  3. Review existing designs in docs/plans/ for related prior work
  4. Identify constraints from the existing architecture
  5. Note what you already know — do NOT re-ask the user for information you can discover

Context Sources Priority

SourceWhat to ExtractPriority
Memory filesStack, conventions, past decisionsHighest
CLAUDE.mdProject structure, hard gatesHigh
Existing codeCurrent patterns, architectureHigh
Recent commitsDirection of developmentMedium
docs/plans/Prior designs and decisionsMedium

STOP — Do NOT proceed to Phase 2 until:

  • Memory files have been checked
  • Relevant codebase areas have been explored
  • You can summarize what you already know about this domain

Phase 2: Idea Exploration

Goal: Understand the user's intent through focused, one-at-a-time questions.

  1. Ask ONE question per message — never multiple questions
  2. Prefer multiple choice questions when possible
  3. Focus on understanding purpose, constraints, and success criteria
  4. Build on context loaded in Phase 1 — skip questions you can already answer
  5. Convert vague answers into specific, testable criteria

Question Flow Decision Table

What You Need to KnowQuestion TypeExample
Core purposeOpen-ended"What problem does this solve for the user?"
Scope boundariesMultiple choice"Should this handle: (A) only logged-in users, (B) all users, (C) depends on role?"
Technical constraintsYes/No + follow-up"Does this need to work offline?"
Priority trade-offsForced ranking"Rank these: speed, correctness, simplicity"
Success criteriaMeasurable target"What does 'working' look like? Can you describe the happy path?"
Non-goalsExplicit exclusion"What should we explicitly NOT build in this iteration?"

Question Rules

RuleRationale
One question per messagePrevents cognitive overload
Multiple choice preferredFaster to answer, reduces ambiguity
Research before askingRespect user's time
Build on memoryDo not re-ask known things
Testable outcomesVague success criteria lead to vague designs

STOP — Do NOT proceed to Phase 3 until:

  • You understand the core purpose and who it serves
  • You know the constraints (technical, timeline, scope)
  • You have identified success criteria and non-goals
  • No critical ambiguities remain about intent

Phase 3: Approach Exploration

Goal: Propose 2-3 distinct approaches with trade-offs and a clear recommendation.

For each approach, present:

SectionContent
NameShort descriptive label
Architecture summary2-3 sentences
Key trade-offsPros and cons
ComplexityLow / Medium / High
RiskWhat could go wrong
Your assessmentWhy you do or do not recommend it

Approach Comparison Template

## Approach A: [Name] (Recommended)
**Summary:** [2-3 sentences]
**Pros:** [list]
**Cons:** [list]
**Complexity:** [Low/Medium/High]
**Risk:** [What could go wrong]
**Why recommended:** [1-2 sentences]

## Approach B: [Name]
...

## Approach C: [Name] (if needed)
...

Lead with your recommended approach and explain why it is the best fit given the constraints discussed in Phase 2.

STOP — Do NOT proceed to Phase 4 until:

  • You have proposed at least 2 approaches
  • Each approach has explicit trade-offs
  • You have made a clear recommendation
  • User has indicated which approach to pursue

Phase 4: Design Presentation

Goal: Present the detailed design in sections, getting approval incrementally.

  1. Present the design in logical sections scaled to complexity
  2. Ask after each section whether it looks right so far
  3. Be ready to go back and revise any section based on feedback
  4. Cover all relevant design dimensions

Design Sections by Complexity

ComplexityRequired SectionsOptional Sections
Simple (1-3 tasks)Architecture, Components—
Medium (4-10 tasks)Architecture, Components, Data Flow, Error HandlingTesting Strategy
Complex (10+ tasks)Architecture, Components, Data Flow, Error Handling, Testing Strategy, Performance, SecurityMigration Plan

Section Presentation Order

  1. Architecture — High-level structure and key decisions
  2. Components — What pieces exist and how they relate
  3. Data Flow — How data moves through the system
  4. Error Handling — What can go wrong and how to handle it
  5. Testing Strategy — How to verify correctness
  6. Performance — Only if relevant constraints exist
  7. Security — Only if handling sensitive data or auth

Present one section at a time. Wait for user confirmation before proceeding.

STOP — Do NOT proceed to Phase 5 until:

  • All relevant design sections have been presented
  • User has approved each section (or revisions were made)
  • The complete design is coherent and addresses all requirements

Phase 5: Documentation and Transition

Goal: Persist the design and hand off to the planning skill.

  1. Write the validated design to docs/plans/YYYY-MM-DD-<topic>-design.md
  2. Commit the design document
  3. Update self-learning memory:

- memory/decisions-log.md — any architectural decisions made - memory/learned-patterns.md — any new conventions discussed

  1. Invoke the planning skill to create a detailed implementation plan

Design Document Template

# [Topic] Design Document

**Date:** YYYY-MM-DD
**Status:** Approved
**Approach:** [Which approach was chosen]

## Problem Statement
[What problem this solves and for whom]

## Design

### Architecture
[High-level structure]

### Components
[Key pieces and their relationships]

### Data Flow
[How data moves through the system]

### Error Handling
[What can go wrong and how it is handled]

## Decisions Made
- [Decision 1]: [Rationale]
- [Decision 2]: [Rationale]

## Non-Goals
- [What was explicitly excluded and why]

## Next Steps
Invoke planning skill to create implementation plan.

STOP — Do NOT proceed until:

  • Design document is saved to docs/plans/
  • Memory files are updated with new decisions/patterns
  • User has confirmed the design is complete

Anti-Patterns / Common Mistakes

Anti-PatternWhy It FailsCorrect Approach
"This is too simple to brainstorm"Every project needs a design, even simple onesDesign can be brief, but must exist
"The user already knows what they want"Users know WHAT, not HOWExplore the HOW through approaches
"I can just start coding"Code without design is technical debt from line 1Design first, code second
"We don't have time to brainstorm"We don't have time to rebuild from poor assumptionsBrainstorming prevents costly rework
"The requirements are clear"Requirements are not design — you still need approachesExplore trade-offs even with clear requirements
Asking 5 questions at onceOverwhelms the user, gets shallow answersOne question per message
Skipping context loadingRe-asks things already knownCheck memory files first
Not proposing alternativesAnchors on first idea, misses better optionsAlways propose 2-3 approaches
Presenting entire design at onceToo much to review, user skimsSection by section with approval
Not persisting decisionsSame discussions repeat in future sessionsUpdate memory files

Anti-Rationalization Guards

If you catch yourself thinking:

  • "Let me just scaffold the project first..." — No. Design first.
  • "The design is obvious..." — Then it will be quick to document. Do it.
  • "The user seems impatient..." — Poor design wastes more time than brainstorming.

Integration Points

SkillRelationshipWhen
self-learningUpstream — provides project context from memoryPhase 1: context loading
planningDownstream — receives approved designPhase 5: transition to planning
task-decompositionComplementary — breaks design into work breakdownWhen design reveals complex scope
spec-writingComplementary — can formalize design into specsWhen formal specifications are needed
verification-before-completionDownstream — verifies design completenessBefore claiming design is done

Concrete Examples

Example: Simple Feature Brainstorm

User: "I want to add dark mode to the app"

Phase 1: [Check memory — React app, Tailwind CSS, no current theme system]
Phase 2: "Should dark mode (A) follow system preference automatically,
          (B) be a manual toggle only, or (C) both with manual override?"
Phase 3: Approach A: CSS variables + Tailwind dark: prefix
         Approach B: Theme context provider with CSS-in-JS
         Recommend A — aligns with existing Tailwind usage
Phase 4: Section 1: Architecture — Tailwind dark mode with class strategy
         Section 2: Components — ThemeToggle component, layout wrapper
Phase 5: Save design doc, invoke planning skill

Example: Transition to Planning

Design approved and saved to docs/plans/2026-03-15-dark-mode-design.md.
Updated memory/decisions-log.md with theme system decision.
Invoking planning skill to create implementation plan.

Key Principles

  • One question at a time — Do not overwhelm
  • Multiple choice preferred — Easier to answer
  • YAGNI ruthlessly — Remove unnecessary features
  • Explore alternatives — Always propose 2-3 approaches
  • Incremental validation — Present and approve in sections
  • Build on context — Use self-learning memory to avoid re-asking known things

Skill Type

RIGID — Follow this process exactly. Every idea goes through all five phases. The design can be brief for simple projects, but it must be documented and approved before any implementation begins.

平台分布

Codex

35.46%
按下载量换算88

Claude

28.22%
按下载量换算70

Cursor

19.77%
按下载量换算49

Gemini CLI

8.31%
按下载量换算21

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills