TCP: TerminalCopilot 🤖⌨️

中文版本

Your AI-powered command line companion. (Actually this README is mostly written by TCP)

Demo

Built with Unix philosophy at its core, you can pass any content by pipeline.

(To be honest, I admit this README is a little showy, and there are many features of TCP that need to be added or improved🥺, but ai feels this project good)

🌟 Features That Will Blow Your Terminal Away

AI Superpowers

• 🧠 Multi-Model Mastery
  -m qwen2.5 | --model deepseek-r1:32b
  Switch between cutting-edge LLMs like changing shells
• 🕵️ Reasoning Mode (-r)
  Two-stage critical thinking:
  Analyze → Execute with different temperature settings
• 💼 PPT Gen Mode (-p)
  Create presentation-ready markdown in terminal
• 🎭 Dual Model Dialogues
  --model2 llama3.1:8b for model vs model debates
• 🛠️ Adding any other features by your own
  Build your personalized AI assistant based on TCP

Terminal Native Magic

• 🛠️ Pipeline Power

  ls -la | tcp -m qwen2.5 "Explain these file permissions"
  man grep | tcp -r "Summarize key flags"

• 🔄 Conversation Mode (-t)

  tcp -t "Debug this Python script" 5

• 📜 Context-Aware
  Maintains session history like your favorite shell

🚀 Installation

Prerequisites: Python 3.9+, Ollama running

# Clone with speed
git clone https://github.com/bingyang-lei/TerminalCopilot.git && cd TerminalCopilot

# Install dependencies (virtualenv recommended)
pip install -r requirements.txt

# Start your AI engine
ollama serve &  # Keep running in background

# Pull models you like
ollama pull qwen2.5 deepseek-r1:32b

# (Optional) set alias
echo 'alias tcp="python /yourpath/main.py"' >> ~/.bashrc
source ~/.bashrc

📚 Usage

Command Parameters

usage: main.py [-h] [-m MODEL] [--model2 MODEL2] [-r] [-p] [-t PROMPT ROUNDS]

options:
  -h, --help            show this help message and exit
  -m MODEL, --model MODEL
                        Designate a model to use, default is qwen2.5
  --model2 MODEL2       (option) Choosing another model to talk with final model
  -r, --reasoning       Enable reasoning mode(more powerful but slower)
  -p, --ppt             enter PPT generation mode
  -t PROMPT ROUNDS, --talk PROMPT ROUNDS
                        Start dialogue mode with initial prompt and max rounds

💡 Why tcp?

Unix Philosophy Embodied

# Chain with classic tools
find . -name "*.py" | tcp -r "Analyze code patterns"
netstat -tulpn | tcp "Explain these network connections"
man sh | tcp "Give me a more friendly manual"

---