explaingit

cancelgumu/webot

17PythonAudience · generalComplexity · 3/5LicenseSetup · moderate

TLDR

A Windows bot that monitors WeChat group chats, summarizes missed messages on request, answers questions when mentioned, and can join conversations automatically using DeepSeek or Claude as its AI backend.

Mindmap

mindmap
  root((webot))
    What it does
      Summarize missed chats
      Answer when mentioned
      Auto join conversations
    AI backends
      DeepSeek
      Claude
    Features
      Pace monitoring
      Group memory
      Web dashboard
    Access method
      Local DB read
      Simulated keyboard
      No WeChat API
    Tech stack
      Python
      WeChat
      Windows
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Automatically summarize what you missed in a busy WeChat group by asking the bot for a recap.

USE CASE 2

Get AI answers to questions asked in a WeChat group without leaving the app or switching to another tool.

USE CASE 3

Monitor conversation pace in a group and have the bot interject automatically when the moment is right.

Tech stack

PythonDeepSeekClaude API

Getting it running

Difficulty · moderate Time to first run · 30min

Windows-only, macOS support is experimental and requires extra steps to extract the WeChat database encryption key.

Use, copy, modify, and distribute freely for any purpose including commercial use, with the original copyright notice kept.

In plain English

webot is an AI-powered bot for WeChat group chats, running on Windows. It monitors selected group conversations, summarizes missed messages on request, answers questions when @mentioned, and can automatically join conversations based on the current chat pace. The README is written entirely in Chinese. The bot reads WeChat message history directly from the app's encrypted local database without connecting to WeChat's servers through any external API. When sending a reply, it simulates keyboard input to type into the WeChat window rather than using any protocol-level access. The README notes this approach carries no account ban risk because no protocols are reversed and no code is injected into the WeChat process. You can configure the bot to use either DeepSeek or Claude as its AI backend. When summarizing a chat, it collects all messages since your last message and passes them to the AI, which returns a structured summary organized by topic, key people, and timeline. For very long conversations, the summary is split and processed in sections to stay within AI context limits. When @mentioned, the bot attaches the last ten minutes of group chat as context so its answer stays relevant to the current discussion. The proactive participation feature monitors how fast messages arrive (with five named pace levels from silent to very active) and decides when to interject automatically. It also builds a long-term memory of the group's personality, recurring jokes, and atmosphere over time, and goes quiet automatically when members are discussing serious negative events. A built-in web dashboard shows real-time metrics, lets you adjust settings, and includes a prompt sandbox for testing AI responses without sending anything to the real group. The bot can be downloaded as a single Windows EXE or run from Python 3.10 source. Experimental macOS support is included but requires extra steps to extract WeChat's database encryption key. The project is MIT-licensed with 17 stars and a clear disclaimer that it is for personal use only.

Copy-paste prompts

Prompt 1
Help me set up webot with Claude as the AI backend on Windows. Walk me through the installation steps and where to put my API key.
Prompt 2
I want webot to summarize a WeChat group chat that has 200 messages since my last read. How does it handle messages that exceed the AI context limit?
Prompt 3
Configure webot proactive participation so it only jumps into the conversation at the two busiest pace levels and stays silent during serious topics.
Prompt 4
I want to access the webot web dashboard. Where does it run and how do I adjust the AI response settings from there?
Open on GitHub → Explain another repo

← cancelgumu on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.