kaixindelele/wechat_agent_bridge_skills
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
Who wouldn't want to use WeChat to control an agent for work? Who wouldn't want to directly verify the authenticity of images on WeChat?
Overview
With GPT Image 2 enabling anyone to generate indistinguishable social media screenshots with one click, many people need an accessible fact-checking tool— No need to install a new app or open a webpage—just upload the abnormal images you see in the group from the most frequently used **, wait a few minutes, and get a report with the original link and authoritative source cross-validation—how delightful! Originally, news fact-checking was just a simple skill prompt, but for convenience in WeChat, I open-sourced the previously simple modified Cursor CLI and Codex CLI with the WeChat Clawbot bridge. This project will connect with and bridge the platform, allowing you to directly drive local AI Agents within WeChat conversations—not only to check news but also to write code, manipulate files, and perform anything an agent can do. Default recommendation: In WeChat fact-checking scenarios, Cursor's WebSearch is better suited for original post targeting and multi-source cross-verification; **More suitable for scenarios emphasizing sandbox permission control and automated execution. Let's assume that users of this project have already deployed Codex or Cursor's CLI on their systems and currently haven't encapsulated the bridge for Claude code. If you install CC, you can have CC parody wrapping a package with all the context provided, which is very fast. Deployment for this project is also very simple. In your agent, enter: Please deploy https://github.com/kaixindelele/wechat_agent_bridge_skills - - Installed and logged in to your chosen agent CLI: - : agent --version (confirm CLI is available) - : codex login
Overview (machine translated)
**Make WeChat your universal assistant. **
**Completed: News - Fact-Checking Portal. **
Completed: PDF Paper Automatic Translation (BabelDOC)
To be updated: Open Source Project Recreation - Analysis Assistant
Send the screenshot to the WeChat bot, and within a few minutes, you will receive an in-depth verification report.
With GPT Image 2 enabling anyone to generate indistinguishable social media screenshots with one click, many people need an accessible fact-checking tool—
No need to install a new app or open a webpage—just upload the abnormal images you see in the group, right in the most frequently used WeChat, wait a few minutes, and get an analysis report with the original link and authoritative source cross-validation—wouldn’t that be great?
Originally, news fact-checking was just a simple skill prompt, but for convenience in WeChat, I open-sourced the previously simple modified Cursor CLI and Codex CLI with the WeChat Clawbot bridge.
This project bridges WeChat with Cursor Agent CLI / OpenAI Codex CLI, allowing you to directly drive local AI Agents in WeChat conversations—not only to check news but also to write code, manipulate files, and perform anything an agent can do.
Default recommendation Cursor Edition: In WeChat fact-checking scenarios, Cursor’s WebSearch is better suited for original post localization and multi-source cross-verification; Codex Edition is more suitable for scenarios emphasizing sandbox permission control and automated execution.
Let’s assume that users of this project have already deployed Codex or Cursor’s CLI on their systems and currently haven’t encapsulated the bridge for Claude code. If you install CC, you can have CC parody wrapping a package with all the context provided, which is very fast.
This project is also easy to deploy. In your agent, enter:
Please deploy https://github.com/kaixindelele/wechat_agent_bridge_skills
Then wait for WeChat to scan the code, then activate the backend service.
✨ Core Competencies
| Scenario | Description |
|---|---|
| News Fact-Checking | Send a news screenshot → Agent automatically uses OCR, multi-source search, cross-verification → returns authenticity verification + original post link + authoritative media coverage |
| Remote AI Programming | Issue commands to an Agent in WeChat, and it will read code, edit files, and run tests on your server |
| Image/File Processing | Send images, files, and voice messages to WeChat bots, and the Agent can directly read and process |
| Session Memory | Automatically reconnects context, so you don’t have to redescribe the problem every time |
| Permission Control | Cursor version supports approval of dangerous operations item by item via WeChat; Codex version controls via sandbox mode |
This page was translated automatically. Technical details may differ from the original.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add kaixindelele/wechat_agent_bridge_skills