Skip to content

jiuxiaomt/Auto-wechat-writing

BranchesTags
 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
public
public
 
 
server
server
 
 
styles
styles
 
 
.env.example
.env.example
 
 
.gitignore
.gitignore
 
 
README.md
README.md
 
 
package-lock.json
package-lock.json
 
 
package.json
package.json
 
 

Repository files navigation

公众号写作神器 / Auto-Wechat Writing

中文 | English

中文文档

AI Agent 驱动的一站式内容创作工具。输入主题，直接生成：正文 → 标题摘要 → 封面图。

功能特性

  • 风格化写作：内置写作风格模板，AI 严格遵循风格约束，文章有辨识度
  • 智能标题摘要：基于正文一键批量生成 5 组标题+摘要，供挑选
  • AI 封面生成：提炼关键内容点，生成 Prompt 并调用文生图模型，一键出图
  • 即拿即用：所有文本一键复制，封面图直接下载
  • 流式输出：正文实时生成，打字机效果流畅
  • OpenAI 兼容：支持所有 OpenAI 兼容 API（含 GPT、Gemini、Qwen、GLM、DeepSeek等）
  • API Key 安全：密钥仅在浏览器本地存储，不传给服务器
  • 自由创建写作风格：在 styles/ 目录添加 .txt 文件即可扩展写作风格，无需改代码
  • 轻量架构：无前端框架依赖，纯 HTML + Tailwind CSS

快速开始

要求：Node.js >= 18

# 1. 克隆项目
git clone https://github.com/lengyi2030/Auto-wechat-writing.git
cd Auto-wechat-writing

# 2. 安装依赖
npm install

# 3. 启动服务
npm start

打开浏览器访问 http://localhost:3000

首次使用会自动弹出 API 设置弹窗，填入 API 配置即可。

开发模式（自动重启）：

npm run dev

API 配置说明

配置项 文本模型 图片模型
API URL OpenAI 兼容 API 地址（如 https://open.bigmodel.cn/api/paas/v4/chat/completions） 文生图 API 地址（如 https://ark.cn-beijing.volces.com/api/v3/images/generations）
API Key 你的 API Key 你的 API Key
模型名称 如 gpt-5.1、qwen3.6-plus、deepseek-v3.2 如 doubao-seedream-5-0-260128、wan2.7-image

文本和图片模型可分别配置不同厂商，互不影响。

安全：设置保存在浏览器 localStorage 中，不会传给服务器。

使用流程

  1. 设置 API：首次访问弹出设置弹窗，填入 API 配置
  2. 选择风格：从下拉菜单选择写作风格
  3. 输入主题：输入你想写的主题或大纲
  4. 生成正文：点击生成，AI 按选定风格流式输出文章
  5. 生成标题：正文完成后一键生成 5 组标题+摘要
  6. 生成封面：AI 提炼关键点并生成封面图 Prompt，选择后生成图片

项目结构

Auto-wechat-writing/
├── server/
│   ├── index.js               # Express 服务入口，端口 3000
│   ├── routes/
│   │   ├── article.js         # 正文 SSE 流式生成 (POST /api/article)
│   │   ├── titles.js          # 标题摘要生成 (POST /api/titles)
│   │   └── cover.js           # 封面 Prompt + 图片生成 + 代理
│   │                          #   POST /api/cover/prompts
│   │                          #   POST /api/cover/generate
│   │                          #   GET/POST /api/cover/proxy
│   └── services/
│       ├── ai.js              # AI API 封装（OpenAI/DashScope/火山引擎适配）
│       └── styleLoader.js     # 风格文件加载器（带缓存）
├── styles/                    # 写作风格 Prompt 模板
│   └── 科技媒体评论.txt
├── public/                    # 前端静态文件
│   ├── index.html
│   ├── css/style.css
│   └── js/app.js
├── .env.example               # 环境变量示例
├── package.json
└── README.md

API 接口

方法 路径 说明
GET /api/styles 获取可用写作风格列表
POST /api/article 流式生成正文（SSE）
POST /api/titles 生成 5 组标题+摘要
POST /api/cover/prompts 根据正文生成封面图 Prompt
POST /api/cover/generate 根据 Prompt 生成封面图
GET /api/cover/proxy?url= 代理图片显示（解决跨域）
POST /api/cover/proxy 代理图片下载

添加写作风格

在 styles/ 目录下新建 .txt 文件即可。文件名即为风格名称。文件内容应包含结构化的 Prompt 指令（参考 styles/科技媒体评论.txt）。

无需改代码，重启服务即可生效。

环境变量

变量 默认值 说明
PORT 3000 服务监听端口

技术栈

  • 后端：Node.js + Express
  • 前端：原生 HTML + Tailwind CSS（无框架依赖）
  • AI 接口：OpenAI 兼容协议（自动适配 DashScope、火山引擎等）
  • 通信方式：SSE（Server-Sent Events）流式输出

许可协议

MIT License

English Documentation

An AI Agent-powered, all-in-one content creation tool. Enter a topic and generate: full article → titles & summaries → cover image.

Features

  • Stylized Writing: Built-in writing style templates. AI strictly follows style constraints for distinctive articles
  • Smart Titles & Summaries: One-click batch generation of 5 title + summary pairs based on the article
  • AI Cover Generation: Extracts key points, generates prompts, and calls image generation models
  • Ready to Use: One-click copy for all text, direct download for cover images
  • Streaming Output: Real-time article generation with smooth typewriter effect
  • OpenAI Compatible: Supports all OpenAI-compatible APIs (GPT, Qwen, GLM, DeepSeek, Volcengine, etc.)
  • API Key Security: Keys are stored locally in the browser, never sent to the server
  • Extensible Styles: Add .txt files in the styles/ directory to add new writing styles — no code changes needed
  • Lightweight Architecture: No frontend framework dependencies — pure HTML + Tailwind CSS

Quick Start

Requirements: Node.js >= 18

# 1. Clone the repository
git clone https://github.com/lengyi2030/Auto-wechat-writing.git
cd Auto-wechat-writing

# 2. Install dependencies
npm install

# 3. Start the server
npm start

Open your browser and visit http://localhost:3000

On first use, an API settings dialog will appear automatically. Fill in your API configuration to get started.

Development mode (auto-restart on file changes):

npm run dev

API Configuration

Setting Text Model Image Model
API URL OpenAI-compatible API endpoint (e.g., https://api.openai.com/v1/chat/completions) Image generation API endpoint (e.g., https://ark.cn-beijing.volces.com/api/v3/images/generations)
API Key Your API key Your API key
Model Name e.g., gpt-5.1, qwen3.6-plus, deepseek-v3.2 e.g., doubao-seedream-5-0-260128, wan2.7-image

Text and image models can be configured with different providers independently.

Security: Settings are saved in the browser's localStorage and are never sent to the server.

Usage Workflow

  1. Configure API: On first visit, a settings dialog appears — enter your API configuration
  2. Select Style: Choose a writing style from the dropdown menu
  3. Enter Topic: Input the topic or outline you want to write about
  4. Generate Article: Click generate — AI streams the article in the selected style
  5. Generate Titles: After the article is complete, generate 5 title + summary pairs with one click
  6. Generate Cover: AI extracts key points and generates cover image prompts, then generates the image

Project Structure

Auto-wechat-writing/
├── server/
│   ├── index.js               # Express server entry, port 3000
│   ├── routes/
│   │   ├── article.js         # Article SSE streaming generation (POST /api/article)
│   │   ├── titles.js          # Title & summary generation (POST /api/titles)
│   │   └── cover.js           # Cover prompt + image generation + proxy
│   │                          #   POST /api/cover/prompts
│   │                          #   POST /api/cover/generate
│   │                          #   GET/POST /api/cover/proxy
│   └── services/
│       ├── ai.js              # AI API wrapper (OpenAI/DashScope/Volcengine compatible)
│       └── styleLoader.js     # Style file loader with caching
├── styles/                    # Writing style prompt templates
│   └── 科技媒体评论.txt
├── public/                    # Frontend static files
│   ├── index.html
│   ├── css/style.css
│   └── js/app.js
├── .env.example               # Environment variable example
├── package.json
└── README.md

API Endpoints

Method Path Description
GET /api/styles Get available writing styles
POST /api/article Stream article generation (SSE)
POST /api/titles Generate 5 title + summary pairs
POST /api/cover/prompts Generate cover image prompts from article
POST /api/cover/generate Generate cover image from prompt
GET /api/cover/proxy?url= Proxy image display (bypass CORS)
POST /api/cover/proxy Proxy image download

Adding Writing Styles

Create a new .txt file in the styles/ directory. The filename becomes the style name. The file content should contain structured prompt instructions (see styles/科技媒体评论.txt for reference).

No code changes needed — restart the server to apply.

Environment Variables

Variable Default Description
PORT 3000 Server listening port

Tech Stack

  • Backend: Node.js + Express
  • Frontend: Vanilla HTML + Tailwind CSS (no framework dependencies)
  • AI Interface: OpenAI-compatible protocol (auto-adapts to DashScope, Volcengine, etc.)
  • Communication: SSE (Server-Sent Events) for streaming output

License

MIT License

About

公众号自动写作神器

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • JavaScript 50.4%
  • HTML 44.0%
  • CSS 5.6%