โœจ Open Source ยท MIT License
๐Ÿฆž

Your Personal AI,
Lightweight & Powerful

A minimal AI assistant that connects to any LLM, runs on Telegram & Feishu, searches the web, and remembers everything โ€” in just ~5900 lines of TypeScript.

~5900
Lines of Code
6+
LLM Providers
0
API Keys Required*

Everything you need

A complete AI assistant framework without the bloat

๐Ÿค–

Multi-LLM Support

Works with Claude, GPT-4, Gemini, Groq, and any OpenAI-compatible API. Switch models with a single config change.

๐Ÿ’ฌ

Multi-Channel

Chat via Telegram or Feishu (Lark). Your AI assistant, wherever you are, on any device.

๐Ÿ”

Free Web Search

Built-in web search using DuckDuckGo โ€” no API key required. Just ask and get real-time information.

๐Ÿง 

Persistent Memory

Long-term memory storage plus daily notes. Your assistant remembers what matters.

โšก

Skills System

Extend capabilities with simple Markdown files. No code needed to teach new tricks.

โฐ

Scheduling

Cron jobs and heartbeat checks. Schedule tasks, set reminders, automate workflows.

Works with your favorite LLM
๐ŸŸ  Anthropic Claude
๐ŸŸข OpenAI GPT
๐Ÿ”ต Google Gemini
โšก Groq
๐ŸŒ OpenRouter
โ˜๏ธ AWS Bedrock

Up and running in minutes

Simple setup, no complex configuration. Just install, add your API key, and start chatting.

  • 1

    Install

    Via npm or build from source

  • 2

    Initialize

    Creates config and workspace

  • 3

    Configure

    Add your LLM API key

  • 4

    Chat

    Start talking to your AI

# Install globally
npm install -g miniclawd

# Initialize workspace
miniclawd onboard

# Add API key to ~/.miniclawd/config.json
# "anthropic": { "api_key": "sk-ant-..." }

# Start chatting!
miniclawd agent -m "Hello!"

# Or start Telegram/Feishu gateway
miniclawd gateway

Ready to build your AI assistant?

Open source, MIT licensed. Fork it, extend it, make it yours.