Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计通过

brainstorming头脑风暴

Agent Skill

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。

总安装

624

周安装

25

GitHub Stars

5

下载量

202
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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/adeonir/agent-skills
仓库路径:skills/brainstorming
安装命令:
npx skills add https://github.com/adeonir/agent-skills --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/adeonir/agent-skills --skill brainstorming

简介

用于结构化探索想法,从模糊需求逐步收敛到可行方向,适用于创意生成与决策支持。

  • 支持在 Codex、Claude、Cursor、Gemini CLI 中使用,基于关键词或任务场景检索信息。
  • 采用发现、发散、收敛、捕获四阶段流程,按需加载参考文档指导各环节输出。
  • 通过 GitHub 安装,使用前需确认权限范围与维护状态,注意是否触发联网或文件操作。
  • 建议结合 ultrathink 进行深度分析,提升方案质量与可行性评估准确性。

SKILL.md

Brainstorming

Structured idea exploration from vague to direction.

Workflow

discover --> diverge --> converge --> capture
  ^________________________|  (if no viable direction)

Discover maps the problem space (context, constraints, success criteria). Diverge generates alternatives using structured techniques. Converge evaluates trade-offs and picks a direction. Capture produces the artifact.

Context Loading Strategy

Load only the reference matching the current phase. The workflow always starts with discovery.

  • Start with discovery.md on every trigger
  • Load diverge.md after discovery quality gate passes
  • Load converge.md after diverge produces at least 4 alternatives
  • Load template only during capture (after user approves direction)

Triggers

Trigger PatternReference
Brainstorm, explore ideas, idea explorationdiscovery.md/brainstorming/references/discovery.md)
What should we build, think through thisdiscovery.md/brainstorming/references/discovery.md)
Explore options, compare approachesdiscovery.md/brainstorming/references/discovery.md)

Notes:

  • diverge.md is not a direct trigger. It is loaded after discovery completes.
  • converge.md is not a direct trigger. It is loaded after diverge completes.
  • The full workflow always starts with discovery regardless of trigger.

Cross-References

discovery.md  -------> diverge.md     (context feeds alternative generation)
diverge.md    -------> converge.md    (alternatives feed evaluation)
brainstorming -------> docs-writer    (direction feeds PRD, epic, design doc)
brainstorming -------> spec-driven    (direction feeds feature specification)
brainstorming -------> design-builder (direction feeds visual exploration)

Guidelines

DO:

  • Generate at least 4 alternatives during diverge, including non-obvious options
  • Challenge every alternative with trade-offs during converge
  • Require explicit user approval before capturing the direction
  • Mark unknowns as TBD rather than inventing constraints
  • Stay at the problem-and-direction level; defer implementation choices
  • Suggest the appropriate next skill after capture

DON'T:

  • Generate only 2-3 obvious alternatives (contrasts: push for non-obvious options)
  • Let the user commit to a direction without evaluating trade-offs (contrasts: challenge every alternative)
  • Include implementation details (contrasts: stay at the problem-and-direction level)
  • Proceed past converge without user confirmation (contrasts: require explicit approval)

Output

  • Artifact: .artifacts/brainstorm/{topic}.md
  • {topic} in kebab-case, derived from the brainstorm subject
  • USE TEMPLATE: templates/brainstorm.md
  • Create .artifacts/brainstorm/ if it does not exist

Error Handling

  • No .artifacts/brainstorm/ directory: create it
  • User arrives with a clear solution: ask if they want to explore alternatives first, or redirect to docs-writer
  • No viable direction after converge: loop back to discovery with refined understanding
  • Scope too broad: decompose into sub-problems, brainstorm one at a time
  • User wants to skip diverge: warn that brainstorming value comes from exploring alternatives, suggest docs-writer if they already have a direction

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.97%
按下载量换算71

Claude

31.18%
按下载量换算63

Cursor

18.02%
按下载量换算36

Gemini CLI

9.5%
按下载量换算19

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills