duliangkuan

duliangkuan / fengyun-publish

Public

End-to-end AI ship pipeline for WeChat Official Accounts · 卷「研究 Agent 的云」

agent-sdk ai automation claude-code content-generation
11
2
85% credibility
Found May 29, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is an end-to-end AI-powered newsletter publishing pipeline. It automatically researches trending AI topics, writes articles in a consistent personal voice, runs multi-stage quality checks, generates matching images, and delivers a polished draft directly to a mobile phone for final review and publishing. The entire workflow runs automatically — from topic discovery to a ready-to-publish draft — with the author only needing to do the final click.

How It Works

1
💡 Discover a topic worth writing about

The system searches news feeds, social media, and research sources to find interesting AI topics that people are talking about right now.

2
📊 See which topics are trending

Based on data from thousands of past articles, the system scores each topic by how likely it is to resonate with readers and perform well.

3
✍️ Write a first draft automatically

An AI writing assistant takes the chosen topic and writes a complete article, following a specific writing style and voice the author has trained it on.

4
🔍 Get feedback and make revisions

The draft goes through multiple rounds of review — one checks the structure, another checks the tone, and a third checks if it sounds authentic — with automatic suggestions for improvements.

5
🎨 Add images and design the layout

The system generates matching illustrations for the article and applies a consistent visual style to make it look polished and professional.

6
📱 Review on your phone and publish

When everything looks good, the finished article lands in your phone's draft folder ready for one final review before going live.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is fengyun-publish?

fengyun-publish is an end-to-end AI pipeline for writing and publishing WeChat Official Account articles. Built in Python with Claude Code orchestration, it automates 19 steps from topic selection through research, drafting, multi-track review, layout, image generation, and pushing directly to the WeChat draft box. The entire workflow runs automatically, leaving only final review and publishing to humans.

Why is it gaining traction?

This stands out because it treats article production as an industrial pipeline rather than a prompt chain. The three-track critique system provides structured quality gates that work independently. Physical checkpoint blockers prevent any step from being skipped, and the system includes anti-AI detection measures to prevent formulaic content. The integration with WeChat's official API via a Cloudflare Worker proxy solves the reliability issues that plague direct API calls.

Who should use this?

Solo content creators running WeChat Official Accounts who want to scale output without sacrificing quality. AI newsletter operators looking for proven automation patterns. Developers building content pipelines who want to study how to structure multi-stage AI workflows with real quality gates and feedback loops.

Verdict

At 11 stars with a credibility score of 0.85%, this is a niche but well-documented project. The extensive README and 48 supporting scripts suggest a mature, production-ready system despite low visibility. If you run a WeChat account and want to automate content creation seriously, study this. For general-purpose AI writing tools, look elsewhere.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.