wechat-to-notion

Save a WeChat article to Notion in three steps: fetch → analyze → save.

Configuration Check (Do This First)

Check if the Notion API key is configured:

echo ${NOTION_API_KEY:0:8}...

If missing, tell the user:

> You haven't configured a Notion API key yet:

> 1. Go to https://notion.so/my-integrations → + New integration → copy the key (starts with ntn_)

> 2. Open your Notion database → ... → Connect to → select your integration

> 3. Set the key in your OpenClaw config — do not paste it into chat:

> ```

> openclaw config set skills.entries.wechat-to-notion.NOTION_API_KEY "ntn_xxx"

> ```

> OpenClaw will inject it as NOTION_API_KEY automatically.

⚠️ Never ask the user to send the API key as a chat message — it will be exposed in conversation logs.

Setup (One-time)

Ask if the user has an existing Notion database. If yes, use it directly. If no, ask for a parent page URL and create one:

curl -s -X POST https://api.notion.com/v1/databases \
  -H "Authorization: Bearer $NOTION_API_KEY" \
  -H "Notion-Version: 2025-09-03" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"type": "page_id", "page_id": "<parent_page_id>"},
    "title": [{"type": "text", "text": {"content": "WeChat Articles"}}],
    "properties": {
      "Title": {"title": {}},
      "URL": {"url": {}},
      "Read Time": {"date": {}},
      "Rating": {"select": {"options": [
        {"name": "⭐", "color": "gray"},
        {"name": "⭐⭐", "color": "gray"},
        {"name": "⭐⭐⭐", "color": "yellow"},
        {"name": "⭐⭐⭐⭐", "color": "orange"},
        {"name": "⭐⭐⭐⭐⭐", "color": "red"}
      ]}},
      "Tags": {"multi_select": {}},
      "Notes": {"rich_text": {}}
    }
  }'

> Match field names to the user's language (e.g. Chinese users get Chinese field names).

Workflow

Step 1: Fetch article

python3 {skillDir}/scripts/fetch_wechat.py <wechat_url> > /tmp/wx_article.json

Step 2: Analyze (inline — reason directly, no subprocess)

Use the read tool to load /tmp/wx_article.json. Read the title and text content from blocks, then produce two outputs by reasoning directly:

Keywords (3–5):

• Only extract core concepts: the specific technologies, products, or domain terms that define what this article is actually about
• Omit generic/broad terms (e.g. "AI", "efficiency", "productivity", "tools", "development")
• Comma-separated, preserve original casing

Rating (1–5 stars):

Based on readability and value, give a star rating:

• ⭐ (1): waste of time — clickbait, no substance, or unreadable
• ⭐⭐ (2): below average — padded, shallow, or poorly organized
• ⭐⭐⭐ (3): decent — has useful content but nothing exceptional
• ⭐⭐⭐⭐ (4): good — well-written, actionable, worth bookmarking
• ⭐⭐⭐⭐⭐ (5): excellent — insightful, well-structured, a must-read in its domain

3 stars and above automatically get a "Featured" tag.

Comment (1 sentence, written in the user's language):

Evaluate the article's readability and value, not summarize its content. Focus on:

• Is it well-structured and easy to follow, or rambling and padded?
• Does it deliver actionable insight, or is it surface-level fluff?
• Who would actually benefit from reading this?
• Example: "Well-structured, flows from theory to hands-on smoothly — ideal for devs wanting to get started with MCP (refreshingly no filler)"
• Example: "Clickbait title, buries the lead under three screens of preamble — the core point could fit in a single tweet"
• Keep it under 35 words. Be direct — praise or criticize with specifics, no hedging.

Step 3: Save to Notion

python3 {skillDir}/scripts/save_to_notion.py \
  /tmp/wx_article.json \
  <notion_db_url> \
  <wechat_url> \
  <read_time_iso8601+08:00> \
  "<kw1>,<kw2>,<kw3>" \
  "<comment>" \
  <rating>

• read_time: current time in the user's local timezone as ISO 8601 with offset, e.g. 2026-03-12T14:00:00+08:00
• keywords: comma-separated string
• comment: the single-sentence comment from Step 2
• rating: integer 1–5 (star rating); 3+ automatically adds "Featured" to tags

The script auto-detects field names from the database schema by type (title, url, date, select, multi_select), writes all content blocks in batches of 100, and posts the comment to the Notion Comments panel.

Notes

• Cover image: extracted from og:image meta tag, inserted as the first block
• Rich text (bold/italic), code blocks, and lists are preserved by fetch_wechat.py
• Read time defaults to current system time with local UTC offset if omitted
• The comment appears in the Notion Comments panel, not in the page body
• Field names are language-agnostic — the script maps by type, not by name