AI

Give WeChat chat records directly to AI: WeChatBridge makes WeChat a local AI workflow portal

Give WeChat chat records directly to AI: WeChatBridge makes WeChat a local AI workflow portal If you use WeChat frequently to access […]

23 September, 2026Updated on 23 September, 2026133 Reading15-minute reading
article directories

Give WeChat chat records directly to AI: WeChatBridge makes WeChat a local AI workflow portal

If you often use WeChat for work, you may encounter this problem:

An important chat record is in the WeChat group.

There may be dozens of texts, a few pictures, a few files, and even videos and links. If you want to hand over these content to AI for summary, analysis or continued processing, the traditional way is often:

Copy chat records → organize content → save files → open AI tools → upload attachments → tell AI what to do.

There are many steps, and chat context, pictures, and attachments can easily be lost in the process.

Is there a more natural way?

Select the chat history directly in WeChat and "forward" it to the AI.

This is an open source project WeChatBridge (WeChat Stream) Problems you want to solve.

Project address:GitHub – freestylefly/WeChatBridge

1. WeChat chat records are themselves a very valuable data source

With the development of AI agents, personal knowledge bases, and local AI tools, an increasingly common workflow is:

Give the information generated daily to AI, and then let AI help us organize, analyze and precipitate it.

WeChat is one of the places where many people generate the most information every day.

Needs discussions in the work group, customer feedback, project progress, technical materials, industry information, articles shared by friends... these contents were originally scattered in the chat window.

The problem is not "no information", but:

It is difficult for information to leave WeChat and enter your own AI workflow.

WeChat natively provides the ability to "forward to other applications", but the system's sharing portal is mainly for applications with Share Extension.

What WeChatBridge does can be understood as adding a bridge between WeChat and local AI tools:

微信聊天
   ↓
合并转发
   ↓
WeChatBridge
   ↓
AI Agent / Obsidian / 剪贴板 / 自定义应用

In this way, WeChat chat is no longer just a "chat record", but can become an input source for local AI workflows.

2. How does it work?

WeChatBridge does not work by reading WeChat databases to obtain chat content, nor does it work by injecting WeChat processes.

It leverages a more natural capability provided by macOS WeChat:

Merge and forward chat records.

Starting from macOS WeChat 4.1.13, you can use "merge and forward" after selecting multiple chat records.

WeChat will generate a ZIP file containing chat content, which can include:

  • TXT chat text
  • picture
  • video
  • Other chat attachments

WeChatBridge catches this content through the macOS Share Extension.

The whole process is very simple:

① 微信中多选聊天记录

        ↓

② 点击“合并转发”

        ↓

③ 选择微信流中的目标

        ↓

④ WeChatBridge 接收聊天归档

        ↓

⑤ 转发给 AI / 保存到知识库 / 复制到剪贴板

The most important point is:

Users do not need to open the main WeChatBridge window first.

The target entry can appear directly in the "Forward to Other Applications" menu on WeChat.

3. What it really solves is actually the "AI workflow entrance" problem

From a functional point of view, WeChatBridge is like a WeChat file forwarding tool.

But from a workflow perspective, it solves another problem:

How to make unstructured information in WeChat quickly enter the AI Agent.

For example, you see dozens of discussions in a project group.

Manual sorting may have been required before:

微信
 ↓
复制聊天
 ↓
整理上下文
 ↓
复制图片
 ↓
下载文件
 ↓
打开 AI
 ↓
上传文件
 ↓
输入提示词

With WeChat streaming, it can become:

微信
 ↓
多选聊天
 ↓
合并转发
 ↓
发给 AI Agent

If combined with the scene prompt words, it can even become:

项目群聊天
     ↓
选择“项目复盘”场景
     ↓
AI Agent
     ↓
总结讨论
提取待办
识别风险
整理决策

This is also the interesting part of this project:

It is not trying to be a chat client again, but turning WeChat into an input node for AI workflows.

4. Built-in multiple AI and productivity tool entrances

Currently, the project already provides multiple native entrances.

Including:

  • Codex
  • Claude
  • bean bag
  • Qianwen Office
  • WorkBuddy
  • WeSight
  • Obsidian
  • clipboard
  • custom application

So users can choose goals based on their own workflow.

For example:

Option 1: Hand it over directly to AI

Discuss a question in the WeChat group.

Select Chat History → Merge and Forward → Send to Codex / Claude.

AI can continue to process these chats.

Option 2: Precipitate into Obsidian

If this chat does not need to be processed immediately by AI, but wants to become your long-term knowledge asset, you can choose directly:

Settle into Obsidian.

The project generates Markdown notes while saving the original archive and related attachments.

So a discussion on WeChat can enter your local knowledge base.

Option 3: Enter the clipboard first

If you are not sure which application to hand it to, you can also copy it to the clipboard first and then manually process it by the user.

This is particularly convenient for some temporary workflows.

5. Scenario function: Make "forwarding to AI" more contextual

If you just hand over the chat history to AI, it is not enough.

Because the same batch of chat records may require completely different processing methods in different scenarios.

For example:

project group

Summarize the discussion and extract decisions, risks and to-do items.

customer base

Extract customer needs, issues and follow-up matters.

Technology Exchange Group

Summarize technical solutions and sort out content worthy of further study.

Industry Exchange Group

Extract industry trends, important information and potential opportunities.

WeChatBridge provides a "scenario" mechanism.

Users can save prompt words for different scenarios and specify applicable Agents.

In this way, the chat history itself is just input, and the scene is responsible for telling the AI:

"How should you understand this content?"

This gradually turns simple file forwarding into a more complete AI workflow.

6. There is another very interesting design: Skills

For AI Agent users,SKILL.md It has become a common way to describe abilities.

WeChatBridge also provides skills management capabilities.

Its idea is:

Chat records are responsible for providing data, and Skill is responsible for providing processing capabilities.

For example, during a chat:

  • webpage link
  • video
  • document
  • picture
  • Various information that requires further analysis

If the Agent itself has the corresponding skills, this content can be processed further.

Therefore, the entire workflow can form:

微信聊天
    │
    ├── 文本
    ├── 图片
    ├── 视频
    ├── 文件
    └── 链接
          ↓
      WeChatBridge
          ↓
      场景 + Skill
          ↓
       AI Agent

This is actually not just a "chat history forwarding", but closer to a local AI information portal.

7. Why does the project emphasize "local"?

WeChat chat records themselves have strong privacy attributes.

Therefore, for a tool for processing WeChat content, one of the most concerned issues for users is:

Did it secretly read my WeChat data?

According to the project public documents, WeChatBridge's design boundaries are relatively clear:

  • Do not read WeChat database
  • Do not decrypt WeChat data
  • Do not inject WeChat process
  • Do not modify the WeChat process
  • Chat content comes from merge forwarding performed by users proactively
  • Chat archives, scenes and records are saved locally
  • Share Extension runs in a macOS sandbox without network permissions

The project also states that screen recording permissions are mainly used to identify the chat name in the WeChat title bar; accessibility permissions are used to activate the target application and perform pasting.

This design reflects a very clear idea:

Use the system's disclosure capabilities as much as possible to complete the workflow, rather than obtaining WeChat data through intrusive methods.

This is especially important for local AI users.

8. It actually conforms to the design method of macOS

WeChatBridge is a native macOS project.

The technology stack mainly includes:

  • Swift 6
  • Swift Package Manager
  • macOS Share Extension
  • App Sandbox
  • Sparkle
  • Swift Testing / XCTest

Project requires macOS 14 Sonoma or later.

From the project structure, we can also see a relatively clear module division:

Sources/
├── WeChatBridgeApp/
│   └── 主应用、设置、转发编排、权限管理
│
├── WeChatBridgeCore/
│   └── 批次、场景、归档、剪贴板等核心逻辑
│
└── WeChatBridgeShare/
    └── macOS Share Extension

Resources/
└── 图标、Plist、entitlements、技能

Scripts/
└── 构建、安装、签名、发布

Tests/
└── 测试代码

This structure also shows that projects are not simply "scripted tools" but are organized in the way macOS natively applies.

IX. Installation is also relatively simple

If you are just ordinary users, you can download DMG directly from GitHub Releases.

The project is currently available in a version signed by Developer ID and notarized by Apple.

After the installation is complete, drag the application into "Applications".

If you are a developer, you can also build directly from source code:

git clone https://github.com/freestylefly/WeChatBridge.git
cd WeChatBridge

swift test

CONFIG=release Scripts/make-app.sh

The project also provides development installation and preview scripts.

Therefore, this project also has certain reference value for developers who like to study macOS automation, Share Extension or AI Agent workflows.

10. What is the difference between it and the traditional "chat history export tool"?

If you look at WeChatBridge and traditional chat history export tools together, you will find that the goals of the two are not exactly the same.

Traditional tools usually focus on:

How to save chat records.

WeChatBridge is more concerned about:

After the chat history is saved, where should we go next?

This is also what makes it more interesting.

It breaks down the entire process into several stages:

信息产生
   ↓
微信
   ↓
主动选择
   ↓
聊天归档
   ↓
WeChatBridge
   ↓
┌─────────────┐
│ AI Agent    │
│ Obsidian    │
│ 剪贴板      │
│ 自定义 App  │
└─────────────┘
   ↓
处理 / 分析 / 沉淀

So it's more like a:

The bridge layer of "WeChat → local AI /knowledge base".

11. What does it mean for AI Agent users?

In the past, when we used AI, we often "open the AI and hand over the data to the AI."

But with the development of Agents, the way they work is gradually changing:

AI should get input wherever information is generated.

WeChat is where information is generated.

Obsidian is the place where knowledge accumulates.

Codex, Claude, etc. are places where information is processed.

What WeChatBridge does is connect these links.

From this perspective, it can be understood as a very lightweight local automation node:

       微信
        │
        ▼
  WeChatBridge
     /   |   \
    /    |    \
   ▼     ▼     ▼
 AI    知识库   其他 App

It does not need to redefine WeChat, nor does it need to replace AI.

It just connects several links that were originally separated.

12. A direction worthy of attention: personal AI workflow

The more noteworthy thing about WeChatBridge may not be that it can transfer WeChat to Claude.

This ability itself is not complex.

What's really worth paying attention to is the workflow behind it:

Let the information generated by individuals every day enter their own AI and knowledge systems at low cost.

Imagine:

In the morning, an industry news item appeared in the WeChat group.

You don't need to copy links or reorganize them.

Forward directly.

AI automatically reads the context and tells you:

  • What does this message say?
  • why it should be followed
  • What does it have to do with past information?
  • Whether further research is needed

In the afternoon, a technical issue was discussed in the project group.

Forward it again.

AI helps you:

  • wrap-up discussion on
  • Draw conclusions
  • Organize TODO
  • Document key decisions

In the evening, add important content to Obsidian.

Over time, WeChat, an originally highly closed information environment, can become an entry point for personal AI workflows.

XIII. Current status of the project

Judging from the public information in the warehouse, WeChatBridge already has a relatively complete product form:

  • macOS 14+
  • Swift 6
  • WeChat native merged forwarding access
  • Multiple AI Agent entrances
  • Obsidian Local Archive
  • Scene prompt words
  • Skill management
  • custom application
  • local history
  • Chinese and English interface
  • Universal 2 Build
  • Developer ID signature and notarization
  • DMG released

The project uses MIT License to open source.

The roadmap also includes Homebrew Cask, as well as more community scenarios and skill packs.

Conclusion: WeChat is just an entrance, and the real goal is information flow

If you only regard WeChatBridge as a "WeChat chat record forwarder", you may underestimate its significance.

What's really interesting about it is:

Release the information in WeChat from the closed chat window and connect it to AI Agents, personal knowledge bases and local automated workflows.

Past:

The message on WeChat → It passed after reading it.

Now:

Information in WeChat → AI processing → knowledge base precipitation → become personal long-term assets.

What WeChatBridge does is the bridge in the middle.

For those who are using AI Agents, Obsidian, and local automation tools, this idea of "directly entering the AI workflow from where the information is generated" may be more worthy of attention than simply adding a new AI client.

Information should not just stay in the chat window.

It can also be an input for the next thinking, the next decision, and the next AI workflow.

Project address:
freestylefly/WeChatBridge

License: MIT License

System requirements: macOS 14 Sonoma or later