wewrite-stats

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

wewrite-stats — 文章数据复盘

wewrite-stats — Article Data Review

运行约定

Operating Conventions

CLI：确定性操作走
```
wewrite
```
命令（需在 PATH；缺失则引导
```
uv tool install wewrite
```
，或在仓库里
```
bash install.sh
```
）。
{home}：用户状态目录 =
```
$WEWRITE_HOME
```
或
```
~/.wewrite
```
（
```
wewrite home
```
可查）。config/style/history/playbook/output/exemplars 全在 {home}，不在仓库；references 文档中的状态路径同此约定。
读取: <路径>
= 用文件读取工具真实读完该文件再继续，不是注释。
references/：本 skill 自带
```
{skill_dir}/references/
```
；references 文档内的
```
{skill_dir}
```
即本 skill 目录。

CLI: Deterministic operations use the
```
wewrite
```
command (must be in PATH; if missing, guide users to run
```
uv tool install wewrite
```
, or run
```
bash install.sh
```
in the repository).
{home}: User state directory =
```
$WEWRITE_HOME
```
or
```
~/.wewrite
```
(can be checked with
```
wewrite home
```
). config/style/history/playbook/output/exemplars are all located in {home}, not in the repository; state paths in references documents follow the same convention.
读取: <路径>
= Actually read the entire file using a file reading tool before proceeding, this is not a comment.
references/: This skill comes with
```
{skill_dir}/references/
```
;
```
{skill_dir}
```
in references documents refers to the directory of this skill.

执行

Execution

读取: {skill_dir}/references/effect-review.md

按其流程执行：

fetch_stats.py --days 7

拉数据 → 匹配并回填

history.yaml

的 stats 字段 → 分析最好/最差表现及原因 → 给出后续选题/标题/框架的调整建议。

前置：需要 config.yaml 里的微信 API 凭证。缺凭证 → 告知用户"数据复盘需要配置公众号 API（config.yaml），当前只能基于 history.yaml 已有记录做定性分析"，然后就 history.yaml 现有内容能分析多少分析多少。刚发布的文章 → 告知等 24h 后再看。

下游影响：回填的 stats 会被选题模块（wewrite-topic）读取——哪种框架/增强策略表现好会加权到下次推荐。这是数据闭环的一半，另一半是 wewrite-learn 的改稿飞轮。

读取: {skill_dir}/references/effect-review.md

Execute according to its process: run

fetch_stats.py --days 7

to pull data → match and backfill the stats field of

history.yaml

→ analyze the best/worst performing content and their reasons → provide adjustment suggestions for future topic selection, titles, and frameworks.

Prerequisites: WeChat API credentials in config.yaml are required. If credentials are missing → inform the user that "data review requires configuring the Official Account API (config.yaml), currently only qualitative analysis can be performed based on existing records in history.yaml", then analyze as much as possible based on the existing content of history.yaml. For newly published articles → inform the user to check again after 24 hours.

Downstream Impact: The backfilled stats will be read by the topic selection module (wewrite-topic) — frameworks/enhancement strategies with good performance will be weighted higher in the next recommendation. This is half of the data closed loop; the other half is the content revision flywheel of wewrite-learn.