review-plan
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChinesePlan Review
计划评审
Comprehensive review of implementation plans using parallel specialized agents.
利用并行的专业Agent对实施计划进行全面评审。
Usage
使用方式
/review-plan # Review plan from current context
/review-plan path/to/plan.md # Review specific plan file/review-plan # 评审当前上下文里的计划
/review-plan path/to/plan.md # 评审指定的计划文件Workflow
工作流程
Step 1: Extract Plan and Check Internal Docs
步骤1：提取计划并检查内部文档
Get the plan to review:
- If path provided, read the file
- Otherwise, use the plan from current conversation context
- Summarize: problem statement, proposed solution, key implementation steps
Check internal documentation:
Use Grep to search for relevant keywords in and files. Look for documented patterns, architectural guidelines, or gotchas related to the plan's area.
docs/*.md获取待评审的计划：
- 如果提供了路径，则读取对应文件
- 否则，使用当前对话上下文里的计划
- 进行总结：问题说明、拟议解决方案、关键实施步骤
检查内部文档：
使用Grep在和文件中搜索相关关键词。查找与计划领域相关的已记录模式、架构指南或注意事项。
docs/*.mdStep 2: Determine Review Scope
步骤2：确定评审范围
Based on plan complexity, decide:
- Simple (single file, minor change): Skip research agent, 2 alternatives
- Medium (few files, new feature): All agents, 3 alternatives
- Complex (architectural, multi-system): All agents + research, 4 alternatives
Do NOT shortcut this workflow:
- "I already know the issues" -- External perspectives find blind spots you can't see
- "This will take too long" -- Parallel agents run simultaneously, the time cost is minimal
根据计划复杂度，决定：
- 简单型（单个文件、微小变更）：跳过研究Agent，提供2种替代方案
- 中等型（少量文件、新功能）：启用所有Agent，提供3种替代方案
- 复杂型（架构级、多系统）：启用所有Agent+研究Agent，提供4种替代方案
请勿简化此工作流程：
- “我已经知道问题所在”——外部视角能发现你看不到的盲区
- “这会花费太多时间”——并行Agent是同时运行的，时间成本极低
Step 3: Spawn Review Agents in Parallel
步骤3：并行启动评审Agent
CRITICAL: Launch agents in a SINGLE message with multiple tool calls.
Do NOT invoke one at a time. Do NOT stop after the first agent.
| Agent | Purpose | Tool |
|---|---|---|
| External Opinions | Get Gemini + Codex input | Skill: |
| Alternatives | Propose 2-4 other solutions | Task: general-purpose |
| Robustness | Check for fragile patterns | Task: general-purpose |
| Adversarial | Maximally critical review | Task: general-purpose |
| Research | Relevant practices online | Skill: |
See references/agent-prompts.md for full prompt templates for each agent.
**关键注意事项：**在单条消息中通过多个工具调用启动Agent。
请勿逐个调用，也请勿在第一个Agent完成后就停止。
| Agent | 用途 | 工具 |
|---|---|---|
| 外部意见Agent | 获取Gemini + Codex的输入 | Skill: |
| 替代方案Agent | 提出2-4种其他解决方案 | 任务：通用型 |
| 健壮性Agent | 检查脆弱性模式 | 任务：通用型 |
| 对抗性Agent | 进行最严苛的评审 | 任务：通用型 |
| 研究Agent | 查找线上相关实践 | Skill: |
每个Agent的完整提示模板请参见references/agent-prompts.md。
Step 4: Synthesize Findings
步骤4：整合评审结果
Collect all agent results and synthesize:
markdown
undefined收集所有Agent的结果并进行整合：
markdown
undefinedPlan Review: {plan_name}
计划评审：{plan_name}
External Opinions
外部意见
Gemini: {summary}
Codex: {summary}
Consensus: {where they agree}
Divergence: {where they disagree}
Gemini: {summary}
Codex: {summary}
共识： {达成一致的内容}
分歧： {存在分歧的内容}
Alternative Approaches
替代方案
| Approach | Key Advantage | Key Disadvantage |
|---|---|---|
| Current plan | {pro} | {con} |
| Alt 1: {name} | {pro} | {con} |
| Alt 2: {name} | {pro} | {con} |
Recommendation: {stick with plan / consider alternative X / hybrid}
| 方案 | 核心优势 | 核心劣势 |
|---|---|---|
| 当前计划 | {优势} | {劣势} |
| 替代方案1: {名称} | {优势} | {劣势} |
| 替代方案2: {名称} | {优势} | {劣势} |
推荐建议： {保留当前计划 / 考虑替代方案X / 混合方案}
Robustness Issues
健壮性问题
Critical (must fix):
- {issue}: {fix}
Warnings:
- {issue}: {fix}
严重问题（必须修复）：
- {问题}: {修复方案}
警告：
- {问题}: {修复方案}
Adversarial Findings
对抗性评审发现
Valid concerns:
- {concern}: {how to address}
Dismissed concerns:
- {concern}: {why it's not a real issue}
有效顾虑：
- {顾虑}: {解决方式}
已排除的顾虑：
- {顾虑}: {为何不构成实际问题}
Research Insights
研究洞察
(if applicable)
- {relevant finding}
（如适用）
- {相关发现}
Revised Plan Recommendations
修订计划建议
{specific improvements to make based on all feedback}
{基于所有反馈提出的具体改进措施}
Changes to Make
待执行变更
- {change 1}
- {change 2}
- {变更1}
- {变更2}
Questions to Resolve
待解决问题
- {unresolved question}
undefined- {未解决问题}
undefinedStep 5: Update Plan
步骤5：更新计划
If significant issues found, offer to revise the plan incorporating the feedback.
如果发现重大问题，建议结合评审反馈修订计划。
Examples
示例
Review a refactor plan -- agents find a robustness issue:
/review-plan
Spawns parallel review agents against the current plan. The robustness agent flags that the migration has no rollback path if it fails midway, and the adversarial agent identifies a race condition under concurrent writes. The synthesis recommends adding a rollback step and a distributed lock.
Review an auth plan with research agent:
/review-plan docs/plans/auth-redesign.md
Reviews the auth redesign plan with all agents including the research agent, which finds that the proposed token rotation strategy has a known edge case documented in the OAuth 2.1 spec. The synthesis recommends adjusting the refresh window based on the research findings.
评审重构计划——Agent发现健壮性问题：
/review-plan
针对当前计划并行启动评审Agent。健壮性Agent指出，迁移过程中如果中途失败则没有回滚路径，对抗性Agent识别到并发写入场景下的竞态条件。整合结果建议添加回滚步骤和分布式锁。
结合研究Agent评审认证计划：
/review-plan docs/plans/auth-redesign.md
启用所有Agent（包括研究Agent）评审认证重构计划，研究Agent发现拟议的令牌轮换策略在OAuth 2.1规范中存在已知的边缘情况。整合结果建议根据研究发现调整刷新窗口。
Troubleshooting
故障排查
Review agents disagree on approach
评审Agent对方案存在分歧
Solution: Focus on the points of consensus first, then evaluate the disagreements by weighing each agent's reasoning against your project constraints. Use the adversarial agent's concerns as a tiebreaker -- if it flags real risk in one approach, prefer the safer alternative.
**解决方案：**先聚焦共识点，再根据项目约束权衡每个Agent的推理来评估分歧。将对抗性Agent的顾虑作为决胜因素——如果它标记某方案存在实际风险，则优先选择更安全的替代方案。
Plan is too vague for meaningful review
计划过于模糊无法进行有意义的评审
Solution: Add concrete details before running the review: specify which files change, what data flows through the system, and what the failure modes are. Agents produce generic feedback when the plan lacks implementation-level specifics.
**解决方案：**在运行评审前添加具体细节：指定要修改的文件、系统中的数据流以及故障模式。当计划缺乏实施层面的具体信息时，Agent只会生成通用反馈。
Notes
注意事项
- Use the Skill tool for and
second-opinion- do not write slash commands directlyresearch-online - External opinions provide model diversity (Gemini + Codex)
- The adversarial agent should be harsh - that's its job
- Robustness review catches patterns that "work in testing, fail in prod" - see references/robustness-patterns.md for examples
- Research agent finds relevant practices and known issues online
- Always synthesize all agent results into actionable improvements
- 使用Skill工具调用和
second-opinion——请勿直接输入斜杠命令research-online - 外部意见提供模型多样性（Gemini + Codex）
- 对抗性Agent的评审要严苛——这是它的职责
- 健壮性评审会发现“测试环境可用，生产环境失效”的模式——示例请参见references/robustness-patterns.md
- 研究Agent查找线上相关实践和已知问题
- 务必将所有Agent的结果整合成可执行的改进建议