Import web pages or WeChat Official Account articles directly into WPS Notes,preserving the original text colors, bold, heading inference, and blockquote highlight blocks, with images inserted at their positions.
Now supports X/Twitter tweets and Threads: uniformly through web_fetch to get the currently visible content, and when image links are captured, insert them into the note at their original positions.
Core Principles
Do not write the fetched Markdown as-is into WPS Notes. The current entry point is web_fetch / web_search, and the fetched main content needs to be organized into WPS XML first before being written into the note; images are still inserted separately.
Correct path:
网页 URL
↓
web_fetch / web_search / batch_web_search
↓
网页正文 + 图片链接
↓
整理成 WPS XML
↓
create_note + batch_edit（正文 + 图片占位符写入 WPS）
↓
get_note_outline / read_blocks（找占位符）
↓
insert_image（逐张插图）
Quick Start
# 用户直接给 URL
web_fetch "https://mp.weixin.qq.com/s/xxxxx"
create_note
batch_edit
insert_image
sync_note
# 用户只给标题或模糊描述
web_search "文章标题或关键词"
ask_user（候选链接不唯一时）
web_fetch "确认后的 URL"
create_note
batch_edit
insert_image
sync_note
Comparison of the Two Modes
| Direct Fetch Mode | Search and Locate Mode | |
|---|---|---|
| Entry | User directly provides a URL | User only provides a title / keywords |
| Get body content | web_fetch |
web_search / batch_web_search + web_fetch |
| Output | Write directly into WPS Notes | Write directly into WPS Notes |
| Candidate confirmation | Generally not needed | When there are multiple candidates, use ask_user |
| Images | insert_image Insert at position |
insert_image Insert at position |
| Read back for acceptance | read_note / read_section / read_blocks |
read_note / read_section / read_blocks |
X/Twitter Support Notes
X content may still be affected by dynamic rendering, login state, and thread expansion limitations. The current Skill uniformly relies on web_fetch the visible body text and image links returned, without additionally assuming browser rendering capabilities.
Workflow
X 推文 URL
↓
web_fetch 获取当前可见正文
↓
提取正文 + 图片链接
↓
整理成 WPS XML
↓
写入 WPS 笔记
Notes
| Scenario | Description |
|---|---|
| Public tweets | Prioritize importing the currently visible body text |
| Tweets requiring login | The body text may not be obtainable, and the user must be clearly informed |
| Long Threads | Only import content that the current tool can reliably fetch |
| Video | Only retain visible text and insertable images; the video itself is not guaranteed |
| Retweets / Quote tweets | Based on the main content currently fetched |
html_to_segments Parsing Logic
No longer relies on html_to_segments local scripts, but the processing goal remains unchanged: organize the web page structure into WPS XML, and preserve the original hierarchy and emphasis styles as much as possible.
Web Content → WPS XML Mapping
| Web page structure / Markdown | WPS XML |
|---|---|
**加粗** / bold semantics |
<strong> |
*斜体* / italic semantics |
<em> |
| Level 1 to Level 3 headings | <h1> / <h2> / <h3> |
| Blockquote | <blockquote> |
| Unordered / ordered lists | <ul> / <li> |
| Table | <table> |
| Link | Preserve text and URL in the paragraph |
Color mapping precision
If the scraping result can clearly preserve color semantics, it can be mapped to available styles in the WPS XML; if the scraping result has already lost color information, do not fabricate colors.
section flattening logic
When a web page has multiple levels of nesting, flatten it in final reading order into headings, body text, quotes, lists, tables, and image placeholders, then write it into WPS.
WPS write stability rules
Keep consistent with the stable write rules of existing note-editing tools; the key constraints are as follows:
正文优先用 batch_edit
少量修补可用 edit_block
图片只能用 insert_image
编辑前必须先拿到 block_id
anchor invalidation retry: Before inserting or supplementing an image, if the block_id has changed, first re- get_note_outline, and if necessary then read_blocks for precise positioning.
outline pagination: After importing a long article, if you need to go back and find image placeholders or supplement positions, first re-fetch the latest outline, then read in segments as needed.
prevent duplicate image insertion: When retrying, first read back the current note to confirm whether the image has already been inserted, avoiding duplicate writes.
Directory structure after download
The final output of the current Skill is a WPS note; there is no requirement to preserve the local download directory. After the import is complete, the note should contain at least:
笔记标题
来源信息（原网页链接）
正文结构（标题 / 段落 / 引用 / 列表 / 表格）
按位置插入的图片
WeChat Official Account title extraction priority
Try in order:
web_fetchthe page title returned byweb_searchthe candidate title matched by- The article title explicitly given by the user in the request
- If all of the above fail → use the link or "Untitled Article" as a fallback
Exception handling
| Scenario | Handling method |
|---|---|
web_fetch Unable to scrape the body text |
Inform the user that the current link cannot be imported, and do not create an empty note |
| Incomplete body text | Allow degraded import of the part that was scraped, and clearly state the missing content |
| Image download / insertion failed | Skip that image, continue importing the body text, and report the number skipped |
| block_id invalid | Re-fetch the outline and target block before continuing |
| Placeholder not found | Read back the body text and relocate; if still not found, skip that image and explain |
| X/Twitter content not visible | Inform the user that the current content cannot be publicly accessed or the thread is not fully expanded |
| Search results not unique | Use ask_user to let the user confirm the link |
Dependencies
| Tool | Purpose |
|---|---|
web_fetch |
Scrape web page body text |
web_search / batch_web_search |
Locate a web page when an exact link is missing |
create_note |
Create the target note for import |
get_note_outline |
Get block_id and the latest structure |
batch_edit / edit_block |
Write the body text and small patches |
insert_image |
Insert images by position |
read_note / read_section / read_blocks |
Read back for acceptance |
sync_note |
Sync after import |
ask_user |
Confirm candidates when the link is unclear |