chestnut-style-analyzer

文风分析

Purpose

This skill turns a creator's historical writing into a reusable 文风说明.md.

It answers one question:

这篇文案像不像这个人？

It is independent, but works best as the first layer of the Chestnut Copy skill set:

1. 文风分析: extract the creator's voice.
2. 爆款文案 SOP: use Zoe/Chestnut's production method.
3. 公众号发布: upload approved drafts to the WeChat draft box.

Source Rule

Recommended source: WeChat Official Account historical articles, because many Chinese creators use WeChat as the mother-draft archive.

Also support local writing samples:

• Markdown
• HTML
• TXT
• JSON exports

Keep voice separate from publishing format. Symbols such as ▼, section spacing, cover style, title hierarchy, and HTML layout belong to the layout layer, not the voice layer.

Credential Setup

Only needed for live WeChat fetching. If the user uses local files, no credential is needed.

Use persistent local credentials by default. The script reads:

1. WECHAT_MP_ENV_FILE
2. ./.env
3. ./.secrets/wechat-mp.env
4. the same files inside this skill directory
5. ~/.wechat-mp.env

Existing environment variables win over file values.

Never commit AppID, AppSecret, access tokens, downloaded article JSON, or generated private style guides.

First Live API Check

Before a real WeChat API call from a new network, get the current IP:

curl -s https://api.ipify.org

Tell the user to add it in:

公众号后台 -> 设置与开发 -> 基本设置 -> IP 白名单

If the API returns 40164, the current IP is not allowlisted.

Fetch Writing Samples

Fetch latest 20 published WeChat articles:

python3 -B scripts/fetch_wechat_articles.py \
  --source published \
  --count 20 \
  --output "${STYLE_ANALYZER_ARTICLES:-./outputs/articles.json}"

Fetch older articles:

python3 -B scripts/fetch_wechat_articles.py \
  --source published \
  --offset 20 \
  --count 20 \
  --append "${STYLE_ANALYZER_ARTICLES:-./outputs/articles.json}" \
  --output "${STYLE_ANALYZER_ARTICLES:-./outputs/articles.json}"