Skip to content

Commit 199f4e2

Browse files
wimi321wimi321
committed
Sharpen README conversion flow
1 parent f6759fc commit 199f4e2

4 files changed

Lines changed: 182 additions & 16 deletions

File tree

‎README.md‎

Lines changed: 75 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -2,20 +2,28 @@
22

33
[English](./README.md) | [简体中文](./README.zh-CN.md)
44

5+
<p align="center">
6+
<img src="./assets/hero-banner.svg" alt="Task Bundle hero banner" width="100%" />
7+
</p>
8+
9+
<p align="center"><strong>Turn AI coding runs into portable, replayable, benchmark-ready task bundles.</strong></p>
10+
<p align="center">The missing middle layer between raw chat logs and heavyweight benchmark platforms.</p>
11+
<p align="center">
12+
<a href="#quickstart"><strong>Quick Start</strong></a> ·
13+
<a href="#real-bundles"><strong>Real Output</strong></a> ·
14+
<a href="./docs/bundle-format.md"><strong>Bundle Format</strong></a> ·
15+
<a href="./ROADMAP.md"><strong>Roadmap</strong></a> ·
16+
<a href="./docs/branding.md"><strong>Brand Assets</strong></a>
17+
</p>
18+
519
[![CI](https://github.com/wimi321/task-bundle/actions/workflows/ci.yml/badge.svg)](https://github.com/wimi321/task-bundle/actions/workflows/ci.yml)
620
[![GitHub stars](https://img.shields.io/github/stars/wimi321/task-bundle?style=social)](https://github.com/wimi321/task-bundle/stargazers)
721
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
822

9-
![Task Bundle hero banner](./assets/hero-banner.svg)
10-
11-
Turn AI coding runs into portable, replayable, benchmark-ready task bundles.
12-
13-
The missing middle layer between raw chat logs and heavyweight benchmark platforms.
23+
Task Bundle is a TypeScript + Node.js CLI for teams building agents, evals, coding benchmarks, and reproducible AI workflows.
1424

1525
Package a task once, inspect it later, compare tools on the same starting point, and generate benchmark-style reports from real artifacts.
1626

17-
Task Bundle is a TypeScript + Node.js CLI for teams building agents, evals, coding benchmarks, and reproducible AI workflows.
18-
1927
Why people star it:
2028
- turn one AI coding run into a clean, shareable directory instead of a screenshot, transcript, or loose patch
2129
- compare Codex, Claude Code, Cursor, or internal agents with real metadata, hashes, and outcome fields
@@ -38,6 +46,66 @@ It is intentionally not:
3846
- a benchmark platform
3947
- a token-by-token recorder
4048

49+
<a id="quickstart"></a>
50+
51+
## Quick Start
52+
53+
Run the repo against real example bundles in about a minute:
54+
55+
```bash
56+
npm install
57+
npm run build
58+
npm run dev -- compare ./examples/hello-world-bundle ./examples/hello-world-bundle-claude
59+
```
60+
61+
If you want the shortest possible proof that the project already works, this is it.
62+
63+
<a id="real-bundles"></a>
64+
65+
## See It On Real Bundles
66+
67+
Inspect a bundle:
68+
69+
```text
70+
$ npm run dev -- inspect ./examples/hello-world-bundle
71+
Task Bundle
72+
-----------
73+
Title: Fix greeting punctuation
74+
Tool: codex
75+
Model: gpt-5
76+
Status: success
77+
Score: 0.93
78+
Workspace files: 1
79+
Events: 3
80+
```
81+
82+
Compare two tools on the same task:
83+
84+
```text
85+
$ npm run dev -- compare ./examples/hello-world-bundle ./examples/hello-world-bundle-claude
86+
Task Bundle Comparison
87+
----------------------
88+
Left tool: codex
89+
Right tool: claude-code
90+
Left score: 0.93
91+
Right score: 0.89
92+
Score delta: 0.04
93+
Workspace file delta: 0
94+
Event count delta: -1
95+
```
96+
97+
Generate a benchmark-style summary from a directory of runs:
98+
99+
```text
100+
$ npm run dev -- report ./examples --out ./dist/benchmark-report.md
101+
Bundles: 2
102+
Average score: 0.91
103+
104+
Ranking
105+
1. Fix greeting punctuation | codex / gpt-5 | success | score 0.93
106+
2. Fix greeting punctuation | claude-code / claude-sonnet-4 | success | score 0.89
107+
```
108+
41109
## Why It Matters
42110

43111
Most AI coding work disappears into screenshots, transcripts, or one-off patches.

‎README.zh-CN.md‎

Lines changed: 76 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -2,20 +2,28 @@
22

33
[English](./README.md) | [简体中文](./README.zh-CN.md)
44

5+
<p align="center">
6+
<img src="./assets/hero-banner.svg" alt="Task Bundle hero banner" width="100%" />
7+
</p>
8+
9+
<p align="center"><strong>把 AI coding 过程变成可分享、可重跑、可比较、可做 benchmark 的任务包。</strong></p>
10+
<p align="center">它正好补上“聊天记录太散、benchmark 平台太重”之间缺失的那层基础设施。</p>
11+
<p align="center">
12+
<a href="#quickstart"><strong>快速开始</strong></a> ·
13+
<a href="#real-bundles"><strong>真实输出</strong></a> ·
14+
<a href="./docs/bundle-format.zh-CN.md"><strong>格式说明</strong></a> ·
15+
<a href="./ROADMAP.zh-CN.md"><strong>路线图</strong></a> ·
16+
<a href="./docs/branding.zh-CN.md"><strong>品牌素材</strong></a>
17+
</p>
18+
519
[![CI](https://github.com/wimi321/task-bundle/actions/workflows/ci.yml/badge.svg)](https://github.com/wimi321/task-bundle/actions/workflows/ci.yml)
620
[![GitHub stars](https://img.shields.io/github/stars/wimi321/task-bundle?style=social)](https://github.com/wimi321/task-bundle/stargazers)
721
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
822

9-
![Task Bundle hero banner](./assets/hero-banner.svg)
10-
11-
把 AI coding 过程，变成可分享、可重跑、可比较、可做 benchmark 的任务包。
12-
13-
它是“原始聊天记录”和“重型 benchmark 平台”之间，那层真正好用的中间层。
23+
Task Bundle 是一个 TypeScript + Node.js CLI，适合 agent、eval、benchmark、可复现实验这类工作流。
1424

1525
一次打包，之后就能 inspect、compare、validate、report，也能把不同工具放到同一起点上做更公平的对照。
1626

17-
Task Bundle 是一个 TypeScript + Node.js CLI，适合 agent、eval、benchmark、可复现实验这类工作流。
18-
1927
大家愿意给它点 star，通常是因为这些点：
2028
- 它能把一次 AI coding 任务整理成干净、稳定、可搬运的目录，而不是散落的截图、聊天记录或 patch
2129
- 它能比较 Codex、Claude Code、Cursor 或内部工具的结果，而且比较依据是真实元数据、哈希和 outcome 字段
@@ -38,6 +46,66 @@ Task Bundle 是一个 TypeScript + Node.js CLI，适合 agent、eval、benchmark
3846
- benchmark 平台
3947
- token 级录制器
4048

49+
<a id="quickstart"></a>
50+
51+
## 快速开始
52+
53+
用仓库自带示例，1 分钟就能跑出一个真实对比：
54+
55+
```bash
56+
npm install
57+
npm run build
58+
npm run dev -- compare ./examples/hello-world-bundle ./examples/hello-world-bundle-claude
59+
```
60+
61+
如果你只想先确认“这项目现在到底能不能用”，这组命令就是最短路径。
62+
63+
<a id="real-bundles"></a>
64+
65+
## 看看真实输出
66+
67+
先 inspect 一个 bundle：
68+
69+
```text
70+
$ npm run dev -- inspect ./examples/hello-world-bundle
71+
Task Bundle
72+
-----------
73+
Title: Fix greeting punctuation
74+
Tool: codex
75+
Model: gpt-5
76+
Status: success
77+
Score: 0.93
78+
Workspace files: 1
79+
Events: 3
80+
```
81+
82+
再比较两个工具在同一个任务上的结果：
83+
84+
```text
85+
$ npm run dev -- compare ./examples/hello-world-bundle ./examples/hello-world-bundle-claude
86+
Task Bundle Comparison
87+
----------------------
88+
Left tool: codex
89+
Right tool: claude-code
90+
Left score: 0.93
91+
Right score: 0.89
92+
Score delta: 0.04
93+
Workspace file delta: 0
94+
Event count delta: -1
95+
```
96+
97+
最后从一组 bundle 直接生成 benchmark 风格摘要：
98+
99+
```text
100+
$ npm run dev -- report ./examples --out ./dist/benchmark-report.md
101+
Bundles: 2
102+
Average score: 0.91
103+
104+
Ranking
105+
1. Fix greeting punctuation | codex / gpt-5 | success | score 0.93
106+
2. Fix greeting punctuation | claude-code / claude-sonnet-4 | success | score 0.89
107+
```
108+
41109
## 为什么值得关注
42110

43111
很多 AI coding 结果最后只留下截图、聊天记录或者一个 patch，后续几乎没法稳定比较。
@@ -73,7 +141,7 @@ task-bundle/
73141
- [docs/bundle-format.md](./docs/bundle-format.md)
74142
- [docs/design-decisions.md](./docs/design-decisions.md)
75143
- [docs/replay-contract.md](./docs/replay-contract.md)
76-
- [docs/branding.md](./docs/branding.md)
144+
- [docs/branding.zh-CN.md](./docs/branding.zh-CN.md)
77145

78146
## 五分钟演示
79147

‎docs/branding.zh-CN.md‎

Lines changed: 28 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,28 @@
1+
# 品牌素材
2+
3+
Task Bundle 在 `assets/` 目录下提供了一套可直接用于仓库展示的视觉素材，让 README、GitHub 首页和分享卡片能保持统一气质。
4+
5+
## 文件说明
6+
7+
- `assets/hero-banner.svg`
8+
中英文 README 顶部使用的主视觉横幅，可继续编辑。
9+
- `assets/social-preview.svg`
10+
GitHub 社交预览图的可编辑源文件。
11+
- `assets/social-preview.png`
12+
已导出的上传版本，适合直接放到 GitHub 仓库设置里。
13+
14+
## 推荐设置
15+
16+
1. 打开仓库设置页。
17+
2. 进入 `General` -> `Social preview`。
18+
3. 上传 `assets/social-preview.png`。
19+
20+
## 重新导出 PNG
21+
22+
在 macOS 上可以直接运行：
23+
24+
```bash
25+
sips -s format png ./assets/social-preview.svg --out ./assets/social-preview.png
26+
```
27+
28+
仓库里保留 SVG，是为了让这套素材更容易继续修改、做版本对比，也更适合长期维护。

‎package.json‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -36,7 +36,9 @@
3636
"templates",
3737
"LICENSE",
3838
"README.md",
39-
"README.zh-CN.md"
39+
"README.zh-CN.md",
40+
"ROADMAP.md",
41+
"ROADMAP.zh-CN.md"
4042
],
4143
"scripts": {
4244
"build": "tsc -p tsconfig.json",

0 commit comments

Comments
 (0)