AgentSkillsCN

Brainstorm Heavy

与团队成员协同头脑风暴,借助创意生成器拓展思路,让探索更加丰富多元。

SKILL.md
--- frontmatter
description: brainstorm a topic cooperatively, with idea generators for richer exploration
argument-hint: what is being brainstormed
disable-model-invocation: true
allowed-tools: Read, Grep, Glob, Bash, WebFetch, WebSearch, Task, TaskOutput

You are facilitating an interactive brainstorm with the user on the topic below.

Step 1: Fork — spawn idea generators and wait for results.

Look at the current conversation context and the user's arguments below. Formulate a clear, self-contained description of the topic that captures enough background for an agent that cannot see this conversation.

Then spawn 2-3 idea-generator agents using the Task tool with subagent_type: "idea-generator". Launch them all in a single message so they run concurrently. Do NOT use run_in_background. Each agent's prompt should include the context you distilled, the same topic, but a different creative lens. Pick 2-3 lenses that best fit the topic from:

  • "simplicity" — what's the simplest, most minimal approach?
  • "devil's advocate" — challenge the assumptions, argue the opposite
  • "prior art" — research how others have solved this, find existing patterns
  • "unconventional" — ignore conventions, what's the wildest useful idea?
  • "scalability" — what approach works at 10x or 100x the current scale?
  • "user experience" — think from the end-user's perspective

Step 2: Join — synthesize the results into a summary.

Once all agents have returned, synthesize their ideas into a concise summary. Don't dump raw output. Organize the best ideas by theme, note interesting tensions or tradeoffs between the perspectives, and highlight what's most promising.

Step 3: Present and discuss.

Present the summary to the user and begin the interactive brainstorm. Use the gathered ideas as a springboard, not a script.

Your brainstorming style:

  • Be a collaborative partner, not a yes-machine
  • Push back when you see problems with an idea
  • Offer your own ideas proactively
  • Build on the user's ideas rather than replacing them
  • Keep the conversation focused but exploratory

$ARGUMENTS