wechat-article

Generates publish-ready WeChat articles through author, editor, and reader agent collaboration.

16.4k|2.5k|Updated Apr 7, 2026
One-click install
npx skills add https://github.com/xbtlin/ai-berkshire --skill wechat-article-xbtlin
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: wechat-article
Source: https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/wechat-article
Command: npx skills add https://github.com/xbtlin/ai-berkshire --skill wechat-article-xbtlin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Single-pass AI writing often produces articles that read well to the writer but confuse or bore the actual audience. This Skill enforces a three-agent workflow—author, editor, and reader—so every WeChat article is checked for depth, readability, and real audience comprehension before publication. ## Core Features & Use Cases - Three-Agent Collaboration: An author agent writes a deep first draft, an editor agent refines structure and expression, and a reader agent reviews from the target audience's perspective. - Parallel Research Phase: Launches 2-3 research agents to gather core content, industry background, and competitive comparisons before writing begins. - Paper Figure Extraction: Renders PDF pages at 900+ DPI with pdftoppm and crops figures with PIL, inserting real high-resolution images instead of placeholders. - Use Case: Ask it to write a WeChat article interpreting a new technical paper; it researches the topic, drafts in conversational Chinese with LaTeX formulas and plain-language explanations, runs editorial and reader reviews, then finalizes a publish-ready Markdown file with extracted figures. ## Quick Start Write a WeChat article interpreting the Qwen3 technical report for readers with some technical background, around 3000-4000 words.

Frequently Asked Questions about wechat-article

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a WeChat article with AI agents?

Provide a topic description such as a paper title or theme. The skill runs parallel research agents, then an author agent drafts the article, and editor plus reader agents review it in parallel before a final revision pass produces the publish-ready Markdown file.

How to extract figures from PDF papers for articles?

Render PDF pages at 900 DPI or higher using pdftoppm, locate figure coordinates on a low-resolution preview, then crop with PIL and save at compress_level=1. Each image should be at least 500KB and stored in an assets directory referenced by relative path.

What input formats does the WeChat article skill accept?

It accepts a plain topic description, such as a technical paper title, an industry trend, or an investment question. You can optionally specify target reader profile, article depth, word count, and writing style; defaults are applied if omitted.

Can the article workflow handle mathematical formulas?

Yes. All formulas must use LaTeX format with inline $...$ or display $$...$$ delimiters, and every formula must be followed by a plain-language explanation. Plain-text formulas are explicitly prohibited.

What are the limitations of multi-agent article writing?

The workflow depends on subagent and web search availability in the host environment, so quality drops without research access. It also requires more time and tokens than single-pass generation due to the research, drafting, and dual review stages.