菜单
Balanced reasoning via parallel creative and critical analysis. Spawns both perspectives independently,
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Convert blogs and documentation sites into zettels for a zettelkasten repository using the `zet` MCP. Use when user wants to generate a zettel from a blog post link or an entire documentation site.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Turn the current conversation context into a PRD and submit it as a GitHub issue. Use when user wants to create a PRD from the current context.
Break a plan, spec, or PRD into independently-grabbable tasks using tracer-bullet vertical slices. Use when user wants to convert a plan into tasks, create implementation tasks, or break down work into actionable items.
Go design and engineering guide covering idiomatic naming, interface-driven architecture, consumer-centric API design, error handling, concurrency, and testing. Use when writing new Go code, refactoring existing code, designing packages, reviewing APIs, or making architectural decisions. Also use when user says "write Go code", "design this package", "review this API", "refactor this", "make this idiomatic", or "clean up this Go code".
| name | brainstorm |
| description | Balanced reasoning via parallel creative and critical analysis. Spawns both perspectives independently, |
When given a problem, execute this process:
Read the discussion till now from shared-reasoning.md if it exists. Use that to identify arguments that have already been agreed upon as well as points of contention that need further exploration. Additionally, user may have provided their perspective on this discussion that will help you continue the analysis further.
Spawn BOTH agents simultaneously with parallel calls:
creatives agent with the EXACT original problemdevils-advocate agent with the EXACT original problemCRITICAL: Pass the original problem verbatim to both. Do NOT summarize, interpret, or add context. Each agent must see identical input.
Do not proceed until both agents have returned their complete analysis.
Present findings in this structure:
## Creative Perspective
[Summarize creative-mind's output]
## Critical Perspective
[Summarize critical-mind's output]
### Convergence (High Confidence)
Where both perspectives agree or complement each other.
### Divergence (Needs Resolution)
Where they conflict. Present both sides without picking a winner.
### Blind Spots
What neither perspective addressed.
### Recommendation
Your synthesis. Acknowledge which perspective you're weighting more heavily
and why. Be explicit about uncertainty.
Write all your output to shared-reasoning__<topic>.md where <topic> is the topic of debate. It should be written into the directory that currently contains files related to the topic being discussed.