DeepSeek Harness plugin

dsh-batch-regression

Run one command N rounds and judge by median/distribution instead of a single run

Jump to install

Source facts

Repository
PangYiMing/dsh-batch-regression
Latest update
Aug 13, 2026
Category
Tools & Capabilities
GitHub stars
1
Format
plugin
Catalog evidence
Upstream dsh.bundle evidence
Evidence path
package.json#dsh.bundle
Checked against
0.1.0-rc.8
Upstream check date
2026-08-20

This evidence comes from the upstream catalog. This site has not installed, run, or security-reviewed the plugin.

Install

Start with a prompt that asks an agent to review the GitHub repository and source. Switch to the command if you want to install it yourself.

Copy this prompt into DSH, Codex, or another agent and ask it to review the GitHub repository and source first.

Do not install or run any commands yet. Read this plugin's GitHub repository, README, and relevant source code. Then answer the questions below clearly and directly so I can decide whether it fits my needs:

1. What is this plugin, and what problem does it solve?
2. Who is it for, and what are its typical use cases?
3. How is it used after installation? Include one minimal example.
4. What known limitations or privacy, security, compatibility, or maintenance risks does it have?
5. Give a clear recommendation: recommend, conditionally recommend, or do not recommend, with reasons.

Distinguish statements documented by the repository, inferences from source code, and unknowns. If evidence is insufficient, say so explicitly. Do not guess or simply repeat the README.

GitHub: https://github.com/PangYiMing/dsh-batch-regression
Plugin: dsh-batch-regression
Author: PangYiMing

Check the source files

Read the README and other files from this plugin directory before installing.

File explorer3 files
README.mdSource · read only

dsh-batch-regression

> DeepSeek Harness 插件 · 批量回归 / DSH plugin for batch regression stats

"数据忽高忽低,到底信哪个"——对同一命令跑 N 轮取中位数和分布,用统计而不是单次结果下结论。

"Numbers keep jumping — which one do I trust?" Run the same command N rounds and judge by median/distribution instead of a single result.

真实方法论 Methodology(实测沉淀)

结论说明
取中位数,不是平均平均会被极值拉偏;中位数更稳
至少 5 轮少于 5 轮的统计没意义
差距 > 20% 结论才稳改前/改后对比,差距 >20% 才下结论
控制变量同一台机器、同一时段、关掉浏览器/IDE 等大头进程
承认抖动来源Apple Silicon P/E 核动态调度会让单次耗时翻倍;笔记本连续跑十几分钟会热降频

> 重要教训:cpulimit/nice/taskpolicy 在 Apple Silicon 上实测全失效——消抖靠"多轮取中位数 + 控制变量",不靠限制 CPU。

使用 Usage

ROUNDS=5 METRIC=time ./scripts/runner.sh "node bench.js"
# samples=5  median=1234ms  (min=1102ms max=1987ms)

ROUNDS=10 METRIC=success ./scripts/runner.sh "npm run build"
# PASS=9/10

什么时候用 / 不用

场景
性能数据抖动,取可信中位数✅ 本工具
改前/改后对比耗时分布✅ 本工具
偶现问题看复现率✅ 本工具(METRIC=success
单次就能判定的事❌ 直接跑
定位"哪次提交引入退化"❌ 用 git bisect(见 dsh-bisect-debug
UI 视觉回归❌ 用 dsh-screenshot-diff

执行纪律

1. 取中位数不是平均。 2. 至少 5 轮。 3. 控制变量:同一台机器、同一时段、关掉大头进程。 4. 承认局限:本地耗时受 P/E 核调度和热降频影响,结论用"差距>20%才稳"标注置信度。 5. 中间失败不中断:某轮 fail 记 FAIL 继续,不因单轮失败停。

安装 Install

# 发布到 npm 后
dsh plugin --profile demo add dsh-batch-regression

# 或从 GitHub 安装
dsh plugin --profile demo add github:PangYiMing/dsh-batch-regression

许可证 License

[MIT](./LICENSE)