nemoclaw-maintainer-pr-comparator

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

PR Comparator

PR 比较工具

Picks the merge winner among competing PRs for a single issue. Tier 0 gates eliminate plumbing failures; Tiers 1-2 score correctness and quality; Tier 3 applies deterministic tiebreakers. Degraded mode handles the case where no PR passes gates.

为单个问题从多个竞争PR中选出应合并的最优项。第0层准入检查排除基础流程故障；第1-2层对正确性和质量进行评分；第3层应用确定性平局决胜规则。降级模式用于处理无PR通过准入检查的情况。

Prerequisites

前提条件

```
gh
```
CLI installed and authenticated
A target repository with an issue that has 2+ open PRs

已安装并完成认证的
```
gh
```
CLI
目标仓库中存在一个带有2个及以上开放PR的问题

Repo policy

仓库策略

Defaults assume NemoClaw conventions (security CODEOWNERS, DCO, CodeRabbit,

docs/

directory). For other repos, edit

repo-policy.md

to override.

默认遵循NemoClaw约定（安全CODEOWNERS、DCO、CodeRabbit、

docs/

目录）。对于其他仓库，可编辑

repo-policy.md

来覆盖默认设置。

Workflow

工作流程

Copy this checklist into your response and check off each step:

text

PR Comparison Progress:
- [ ] Step 1: Parse issue (body + comments) for acceptance criteria
- [ ] Step 2: Discover candidate PRs (default-order search with stop conditions)
- [ ] Step 3: Detect supersession (parse PR bodies)
- [ ] Step 4: Run Tier 0 gates per PR
- [ ] Step 5: Run Tier 1 correctness checks per PR
- [ ] Step 6: Run Tier 2 quality checks per PR
- [ ] Step 7: Compute weighted scores
- [ ] Step 8: Apply Tier 3 ranking (happy path or degraded mode)
- [ ] Step 9: Emit verdict using templates/verdict.md

将以下检查清单复制到你的回复中，并勾选完成的步骤：

text

PR Comparison Progress:
- [ ] Step 1: Parse issue (body + comments) for acceptance criteria
- [ ] Step 2: Discover candidate PRs (default-order search with stop conditions)
- [ ] Step 3: Detect supersession (parse PR bodies)
- [ ] Step 4: Run Tier 0 gates per PR
- [ ] Step 5: Run Tier 1 correctness checks per PR
- [ ] Step 6: Run Tier 2 quality checks per PR
- [ ] Step 7: Compute weighted scores
- [ ] Step 8: Apply Tier 3 ranking (happy path or degraded mode)
- [ ] Step 9: Emit verdict using templates/verdict.md

Step 1: Parse issue

步骤1：解析问题

Extract acceptance criteria from issue body and all comments:

bash

gh issue view <issue-number> --json title,body,comments

Read every comment — commenters often add asks the body doesn't capture.

从问题正文及所有评论中提取验收标准：

bash

gh issue view <issue-number> --json title,body,comments

阅读所有评论——评论者常会提出正文未涵盖的需求。

Step 2: Discover candidate PRs

步骤2：发现候选PR

bash

scripts/find-candidates.sh <issue-number>

Applies a single default order with stop conditions.

bash

scripts/find-candidates.sh <issue-number>

应用带有终止条件的单一默认排序规则。

Step 3: Detect supersession

步骤3：检测替代关系

bash

scripts/parse-supersession.sh <pr-number-1> <pr-number-2> ...

Parses each PR body for

supersedes #N

replaces #N

closes in favor of #N

folds in #N

. A PR that supersedes another wins ties immediately.

bash

scripts/parse-supersession.sh <pr-number-1> <pr-number-2> ...

解析每个PR正文，查找

supersedes #N

、

replaces #N

、

closes in favor of #N

、

folds in #N

等标识。若一个PR替代了另一个PR，则在平局时直接胜出。

Step 4: Tier 0 gates

步骤4：第0层准入检查

bash

scripts/collect-gates.sh <pr-number>
scripts/check-coderabbit-threads.sh <pr-number>

Five gates, all mandatory. See

checks/tier-0-gates.md

for the full list and interpretation.

bash

scripts/collect-gates.sh <pr-number>
scripts/check-coderabbit-threads.sh <pr-number>

共5项准入检查，均为必填项。完整列表及说明请查看

checks/tier-0-gates.md

。

Step 5: Tier 1 correctness

步骤5：第1层正确性检查

Six checks, all LLM judgments. See

checks/tier-1-correctness.md

for evidence requirements per check.

共6项检查，均由LLM进行判断。每项检查的证据要求请查看

checks/tier-1-correctness.md

。

Step 6: Tier 2 quality

步骤6：第2层质量检查

Three checks, all LLM judgments. See

checks/tier-2-quality.md

共3项检查，均由LLM进行判断。详情请查看

checks/tier-2-quality.md

。

Step 7: Weighted score

步骤7：加权评分

Each pass = full points
Each yellow = half points
Each fail = zero
Tier 1 weight: 2.0× per check
Tier 2 weight: 1.0× per check

通过：满分
警告：半分
失败：零分
第1层检查权重：每项×2.0
第2层检查权重：每项×1.0

Step 8: Tier 3 ranking

步骤8：第3层排名

Branch on whether any PR passes all Tier 0 gates. See

tiebreakers.md

for happy-path tiebreakers, degraded-mode distance-to-ready ranking, and the behavior-coverage matrix.

根据是否有PR通过所有第0层准入检查进行分支处理。正常流程平局决胜规则、降级模式下的就绪度距离排名以及行为覆盖矩阵请查看

tiebreakers.md

。

Step 9: Emit verdict

步骤9：输出裁决结果

Use

templates/verdict.md

. Every judgment must carry evidence (file:line refs, diff snippets), reasoning chain, and the score it contributed.

使用

templates/verdict.md

模板。每项判断必须附带证据（文件:行号引用、差异片段）、推理链及其对应的评分。

Reference files

参考文件

```
repo-policy.md
```
— configurable defaults per target repo
```
checks/tier-0-gates.md
```
— plumbing gates
```
checks/tier-1-correctness.md
```
— six correctness checks
```
checks/tier-2-quality.md
```
— three quality checks
```
tiebreakers.md
```
— Tier 3 ranking and degraded mode
```
templates/verdict.md
```
— output template
```
validation/backtest.md
```
— backtest the skill against historical cases

```
repo-policy.md
```
— 目标仓库的可配置默认设置
```
checks/tier-0-gates.md
```
— 基础流程准入检查
```
checks/tier-1-correctness.md
```
— 6项正确性检查
```
checks/tier-2-quality.md
```
— 3项质量检查
```
tiebreakers.md
```
— 第3层排名及降级模式
```
templates/verdict.md
```
— 输出模板
```
validation/backtest.md
```
— 基于历史案例对该技能进行回测

Scripts (execute, do not read)

脚本（仅执行，无需阅读）

```
scripts/find-candidates.sh
```
— PR discovery
```
scripts/collect-gates.sh
```
— Tier 0 gate evaluation
```
scripts/check-coderabbit-threads.sh
```
— GraphQL thread resolution
```
scripts/parse-supersession.sh
```
— body parsing for supersession refs
```
scripts/render-verdict.py
```
— verdict scorecard renderer

```
scripts/find-candidates.sh
```
— PR发现
```
scripts/collect-gates.sh
```
— 第0层准入检查评估
```
scripts/check-coderabbit-threads.sh
```
— GraphQL线程解析
```
scripts/parse-supersession.sh
```
— 正文解析以查找替代关系引用
```
scripts/render-verdict.py
```
— 裁决评分卡渲染器

What this skill does NOT do

该技能不支持的功能

These require infrastructure beyond GitHub API + LLM and are deferred to v2 modules:

Running each PR's code against adversarial inputs (sandboxed execution)
Cross-issue regression sweep (separate skill)
Revert simulation against neighbor PRs
Static analyzer integration (CodeQL, Semgrep)

这些功能需要GitHub API + LLM之外的基础设施，将推迟到v2版本模块实现：

在对抗性输入下运行各PR代码（沙箱执行）
跨问题回归扫描（独立技能）
针对关联PR的回滚模拟
静态分析器集成（CodeQL、Semgrep）