菜单
Adversarial brainstorming. Claude and Codex independently research then debate until Nash equilibrium. For solution exploration, feasibility analysis, exhaustive enumeration.
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
Document codebase as-is via parallel sub-agents. Use when researching how code works, mapping architecture, or answering questions about existing systems. Auto-detects thoughts/ directory — if present, integrates historical context and syncs; if absent, skips those steps.
Create the Questions artifact before codebase research, design, planning, or implementation. Use for feature tickets, architecture changes, unclear code work, CRISPY/QRSPI Questions phase, or when the agent needs to define code-answerable unknowns and human questions before researching or planning.
Create a concise Design Concept artifact after questions and codebase research are available. Use for CRISPY/QRSPI Design phase, architecture alignment, feature design discussion, or turning current-state research into a small human-reviewable proposed shape before detailed planning.
Create an interface-first Structure Outline before detailed planning or implementation. Use for CRISPY/QRSPI Structure Outline phase, vertical planning, method signatures, type contracts, API/component boundaries, or converting an accepted design concept into end-to-end slices with verification checkpoints.
Create comprehensive handoff documents for transferring work to another agent session. Use when the user requests to create a handoff, end a session, or document work-in-progress for future continuation.
Create detailed implementation plans through interactive research and iteration. Use when starting a new feature or ticket that needs a comprehensive technical plan. Saves plans to the repo-appropriate plans location and emits Personal Knowledge events when applicable.
| name | codex-brainstorm |
| description | Adversarial brainstorming. Claude and Codex independently research then debate until Nash equilibrium. For solution exploration, feasibility analysis, exhaustive enumeration. |
| allowed-tools | mcp__codex-cli__codex, Read, Grep, Glob |
/codex-review)⚠️ Independent Research → Adversarial Debate → Nash Equilibrium ⚠️
Nash Equilibrium = Neither party can unilaterally change strategy to achieve a better outcome
| Phase | Action | Output |
|---|---|---|
| 1 | Claude independent research + analysis, forms Position A | Claude's optimal hypothesis |
| 2 | Codex independent research + analysis, forms Position B | Codex's optimal hypothesis |
| 3 | Multi-round adversarial debate, mutual attacks | Debate exchange record |
| 4 | Check equilibrium, no further improvements possible | Equilibrium or divergence |
| 5 | Output final report | Nash Equilibrium report |
⚠️ Must let Codex research independently; do NOT feed Claude's analysis results ⚠️
mcp__codex-cli__codex({
prompt: `You are a senior architect. Conduct an **independent analysis** of the following topic.
## Topic
${TOPIC}
## Constraints
${CONSTRAINTS}
## ⚠️ Important: You must research independently ⚠️
Before forming conclusions, you **must** first:
1. Run \`ls src/\` to understand the directory structure
2. Search related code: \`grep -r "keyword" src/ --include="*.ts" -l | head -10\`
3. Read relevant files to confirm existing implementations
## Output Requirements
1. Research summary (related modules, existing patterns)
2. Your position + supporting arguments
3. Potential risks`,
sandbox: 'read-only',
fullAuto: false,
});
Structure per round:
| Condition | Description | Result |
|---|---|---|
| Nash Equilibrium | Neither side can raise new attacks | Output equilibrium |
| Convergence | Both positions converge | Output consensus |
| Max rounds | 5 rounds reached with remaining divergence | Output divergence report |
| File | Purpose |
|---|---|
templates.md | Claude/debate/report templates |
techniques.md | Attack/defense techniques |
equilibrium.md | Equilibrium determination flow |
Input: What implementation approaches are available for this requirement?
Phase 1: Claude independent research → Position A (Solution X is optimal)
Phase 2: Codex independent research → Position B (Solution Y is optimal)
Phase 3: Adversarial debate
- R1: Claude attacks Y's scalability / Codex attacks X's complexity
- R2: Claude rebuts / Codex concedes and updates position
- R3: Both converge to Solution Z, no further attacks → Equilibrium
Phase 4: Output Nash Equilibrium = Solution Z