AI Agent Hub
Back to plugins
⚙️

gzhflow

Workflow Updated 2026.09.02

Run the following command in DeepSeek Harness:

dsh plugin install Andiii208/gzhflow

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Andiii208/gzhflow in DeepSeek Harness to install this plugin. The source repository is available at https://github.com/Andiii208/gzhflow

About this plugin

Writing a WeChat Official Account article with AI is rarely about not being able to produce words. The real pain lies in the chain of tedious steps between an idea and a polished draft: digging up real material, writing in a personal voice, stripping away AI flavor layer by layer, pairing a non-gimmicky image, converting Markdown into the platform HTML template, and safely pushing to the draft box. Manually stitching these together is slow and error-prone, while asking a model for one-shot output leaves quality gates effectively meaningless.

gzhflow solidifies this pipeline into six stages, each gated by at least one quality check: deterministic script validation, LLM self-check checklists, and human review. Voice routing, the de-AI-flavor handbook, the four-stage image prompt compiler, and layout theme switching all live in config files and examples templates, so the workflow itself is decoupled from any individual. Under the hood it is a pure stdlib CLI with zero external dependencies and full cross-platform support.

It works with mainstream agents including Claude Code, Cursor, Codex CLI, Gemini CLI, Qwen Code, and DeepSeek, and is also compatible with the Hermes SKILL.md format. If you are an operations lead, an indie developer, or part of a content team building multi-agent writing pipelines who needs to produce WeChat articles on a regular cadence, gzhflow offers a reusable framework that takes one topic in and delivers a formatted, quality-gated draft out.

Use Cases

  • Turn a topic into a formatted, image-ready, de-ai-flavored WeChat draft in one pipeline run
  • Reuse the same six-stage writing workflow across Claude Code, Cursor, DeepSeek, and other agents
  • Layer deterministic script checks, LLM self-check lists, and human review as quality gates on a content pipeline

Best For

  • Operations leads or indie devs who publish WeChat articles on a regular cadence
  • Content teams building multi-agent collaborative writing pipelines
  • Creators who want to codify personal writing methodology into a reusable framework