mb-review

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

mb-review — Multi-expert Memory Bank review

mb-review — 多专家Memory Bank审查

What it does: runs independent reviewers over architecture, scope, backlog, security, and Memory Bank discipline.
Use it when: you need a clean review gate before execution, after bootstrap, or before trusting the docs.
Input: an existing
```
.memory-bank/
```
.
Output: reviewer reports in
```
.tasks/TASK-MB-REVIEW/
```
plus a synthesized list of fixes and review verdicts.

功能： 针对架构、范围、待办项、安全以及Memory Bank规范运行独立审查程序。
适用场景： 当你需要在执行前、项目启动后，或是确认文档可信任前设置一个清晰的审查关口时使用。
输入： 已存在的
```
.memory-bank/
```
目录。
输出： 存放在
```
.tasks/TASK-MB-REVIEW/
```
目录下的审查报告，以及综合后的修复清单和审查结论。

Goal

目标

Detect gaps, contradictions, broken traceability, and non-compliance early.

及早发现内容缺失、矛盾、可追溯性中断以及不合规问题。

Preconditions

前置条件

```
.memory-bank/
```
exists.

```
.memory-bank/
```
目录存在。

Process

流程

1) Create review task folder

1) 创建审查任务文件夹

Create:

```
.tasks/TASK-MB-REVIEW/
```

创建：

```
.tasks/TASK-MB-REVIEW/
```

2) Spawn reviewers (fresh contexts)

2) 启动审查Agent（全新上下文）

Spawn these subagents in parallel (max 5–7):

Architect —
```
./agents/shared-review-architect.md
```
Scope/RTM —
```
./agents/shared-review-scope.md
```
Plan/backlog —
```
./agents/shared-review-plan.md
```
Security —
```
./agents/shared-review-security.md
```
MBB compliance —
```
./agents/shared-mb-reviewer.md
```
Code quality (conditional: if repo has code) —
```
./agents/shared-review-code.md
```

Each reviewer must:

write a detailed report to
```
.tasks/TASK-MB-REVIEW/
```
return only a short summary + verdict to the orchestrator

并行启动以下子Agent（最多5-7个）：

架构师 —
```
./agents/shared-review-architect.md
```
范围/RTM —
```
./agents/shared-review-scope.md
```
计划/待办项 —
```
./agents/shared-review-plan.md
```
安全 —
```
./agents/shared-review-security.md
```
MBB合规 —
```
./agents/shared-mb-reviewer.md
```
代码质量（触发条件：若仓库包含代码） —
```
./agents/shared-review-code.md
```

每个审查Agent必须：

向
```
.tasks/TASK-MB-REVIEW/
```
目录提交详细报告
仅向编排器返回简短摘要 + 审查结论

3) Synthesize and decide

3) 综合分析并决策

As orchestrator:

combine findings
deduplicate
rank issues P0–P3
produce a concrete fix plan

If the repo is preparing for

/autonomous

treat unresolved P0/P1 issues as blocking
do not allow batch execution until the final verdict is
```
APPROVE
```

作为编排器：

合并所有审查发现
去重
按P0-P3等级对问题进行优先级排序
生成具体的修复计划

如果仓库正在为

/autonomous

模式做准备：

未解决的P0/P1问题视为阻塞项
在最终结论为
```
APPROVE
```
前不允许批量执行

4) Gate

4) 准入校验

If any reviewer returns

REJECT

fix MB
re-run mb-review

如果任意审查Agent返回

REJECT

：

修复MB
重新运行mb-review

Definition of done

完成定义

```
.tasks/TASK-MB-REVIEW/
```
contains the reviewer reports.
Orchestrator produced an actionable prioritized fix list.
Final verdict: APPROVE.

```
.tasks/TASK-MB-REVIEW/
```
目录包含所有审查Agent的报告
编排器已生成可落地的优先级修复清单
最终结论：APPROVE