Lazy Terminal

🚀 Lazy Terminal

Lazy Terminal is an intelligent command-line assistant that connects language models with external tools through the MCP (Modular Component Protocol). It can dynamically determine whether to provide a direct answer or invoke a registered tool based on the context, enabling seamless terminal automation and intelligent interaction (you're lazy; it isn't).

🚀 Quick Start

Installation

pip install -r requirements.txt

Configuration

• Edit the config/models.yaml file and add your model credentials.
• Edit the config/mcp_servers.yaml file and list your MCP servers.

Running

Start the assistant using the following command:

python main.py

Start interacting with the Lazy Terminal directly from the console!

✨ Features

• Dynamic Tool Routing: Automatically choose between direct answers and tool invocations.
• Context Memory: Maintain the conversation context (default is 50 interactions).
• Multi-model Support: Easily integrate various language models through Coagent.
• Robust JSON Handling: Extract and process structured model responses.
• Secure Bash Execution: Securely execute controlled bash scripts through MCP tools.

💻 Usage Examples

Start the bash runner server:

python test_servers/bash_runner.py

Example interaction: User input:

You: List all .txt files in the ~/Documents directory and save the results to files.txt

Assistant tool invocation:

{
  "tool_name": "run_bash_script",
  "parameters": {
    "commands": "ls ~/Documents/*.txt > files.txt"
  }
}

The Lazy Terminal will execute this operation and return the results.

🔑 Environment Variables

Set the model API keys as environment variables, for example:

export DEEPSEEK_API_KEY="your-api-key"

Ensure that these keys match the configuration in models.yaml.

📌 Roadmap

• [ ] Implement advanced memory compression.
• [ ] Add an interactive UI (terminal/web).
• [ ] Enhance error handling and logging.
• [ ] Expand the toolset (e.g., file handling, browser automation).
• [ ] Support multi-tool invocation

🤝 Acknowledgments

• Coagent
• MCP

Enjoy building your intelligent terminal assistant!