
qbu11/wechat-article-spider
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
WeChat public account article crawling CLI tools
Overview
The installer first displays the write schedule and requests confirmation, installing three standard Agent Skills into the Claude Code, Codex, and generic .agents/skills directories. It does not have postinstall and supports --dry-run, project-level installation, and secure uninstallation. Requires Node.js 22.13 or later; Current Node.js 24 LTS recommended. No resident services, no independent databases, no auto-running background tasks. SQLite is a built-in Node.js capability; the CLI automatically creates and maintains it; It is only used to store subscriptions, indexed articles, and synchronized status. All commands write structured JSON to stdout, making it easy to stably call any agent or script. Packaged Skills always use fixed versions of npx calls and do not rely on globally installed raw commands. View full command: By default, query does not fetch each main text but instead queries local indexes and network sources in parallel, returning compact JSON as quickly as possible. --keywords is the cross-official account keyword-search; --account with a clear date boundary is the account-window for a single public account. The remote public account time window is only best-effort; an empty result does not mean the account has not published; To achieve sustainable, repeatable query coverage, you should subscribe to a verified feed, perform sync, and then check the local index. Each result provides originalUrl and discoveryUrl: only links with hostnames strictly mp.weixin.qq.com can be called WeChat original links; Sogou redirects and other hosts can only be used as discovery sources. The complete Skill flowchart and data processing procedures explain intent routing, deduplication, source, and synchronization rules. WeChat official accounts do not have a stable public subscription interface for this purpose. Therefore, this project only persistently subscribes to RSS 2.0, Atom, or JSON feed addresses that you explicitly provide and have been content verified. The public account name or ID ** found is guessed to be a subscription source. - search --scope global Relies on Sogou WeChat search and may encounter captchas, traffic limits, or result changes; The tool will clearly report errors and will not bypass access controls. - read supports public WeChat original links; Deleted, restricted, or verified-needed pages will return the corresponding status. - sync is an explicit operation. When scheduled updates are needed, have your trusted external scheduler call the fixed npx -y @qbu11/[email protected] sync --json.
Overview (machine translated)
Let AI Agents search, read, and subscribe to WeChat official account articles.
Chinese · English
One-Command Installation
Before the first npm version is released, you can also directly use a fixed GitHub version:
The installer first displays the write plan and requests confirmation, installing three standard Agent Skills into the Claude Code, Codex, and generic .agents/skills directories. It does not have postinstall, supports --dry-run, project-level installation, and secure uninstallation.
Requires Node.js 22.13 or later; Current Node.js 24 LTS recommended.
Why is it simple enough?
No resident services, no independent databases, no auto-running background tasks. SQLite is a built-in Node.js capability; the CLI automatically creates and maintains it; It is only used to store subscriptions, indexed articles, and synchronized status.
| Skill | Capabilities | Data source |
|---|---|---|
wechat-search |
Keyword search, specific Official Account time slot search, Discover Official Account | Local index + Sogou WeChat Search |
wechat-article |
Read the main text and metadata from the original link | mp.weixin.qq.com |
wechat-subscribe |
Save subscriptions, sync manually, check health status | RSS 2.0 / Atom / JSON Feed |
All commands write structured JSON to stdout, making it easy to stably call any agent or script.
Get it started in 30 seconds
After installation, you can directly tell the Agent:
Search for recent articles about OpenAI on WeChat public accounts and provide the original link.
Read this WeChat article to summarize the main points and retain the source information.
subscribe to this Atom address, then sync once and list new articles.
You can also use the CLI directly:
Packaged Skills always use fixed versions of npx calls and do not rely on globally installed raw commands. View the full command:
query By default, it does not fetch every main text, but instead queries local indexes and network sources in parallel, returning compact JSON as quickly as possible. --keywords is cross-public account keyword-search; --account with a clear date boundary is for a single public account account-window. The remote public account timewindow is just best-effort; an empty result does not mean the account hasn’t published; To achieve sustainable, repeatable query coverage, you should subscribe to a validated feed, run sync, and then look up the local index. Each result provides originalUrl and discoveryUrl: Only links with hostnames strictly mp.weixin.qq.com can be called WeChat original links; Sogou redirects and other hosts can only be used as discovery sources. The complete Skill flowchart and data processing procedures explain intent routing, deduplication, source, and synchronization rules.
Subscription Boundaries
This page was translated automatically. Technical details may differ from the original.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add qbu11/wechat-article-spider