🐢 Lazy Terminal

Lazy Terminal is an intelligent command-line assistant that connects language models with external tools through MCP (Modular Component Protocol). It dynamically determines whether a query should be directly answered or passed to a registered tool, enabling seamless terminal automation and smart interaction (you are lazy; it isn't).

---

🔧 Features

• Dynamic Tool Routing: Automatically chooses between direct response or tool invocation.
• Contextual Memory: Maintains conversation context (default 50 interactions).
• Multi-model Support: Easily integrate various language models via Coagent.
• Robust JSON Handling: Extracts and processes structured model responses.
• Secure Bash Execution: Runs safe, controlled bash scripts through MCP tools.

---

🚀 Quick Start

Installation

pip install -r requirements.txt

Configuration

• Edit config/models.yaml to include your model credentials.
• Edit config/mcp_servers.yaml to list your MCP servers.

Running

Launch the assistant with:

python main.py

Start interacting with Lazy Terminal directly from your console!

---

🛠 Example Usage

Start the bash runner server:

python test_servers/bash_runner.py

Sample interaction:

User input:

You: List all .txt files in ~/Documents and save output to files.txt

Assistant tool invocation:

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

Lazy Terminal will then execute this safely and return the results.

---

🔑 Environment Variables

Set model API keys as environment variables, e.g.:

export DEEPSEEK_API_KEY="your-api-key"

Ensure they match your settings in models.yaml.

---

📌 Roadmap

• [ ] Implement advanced memory compression.
• [ ] Add interactive UI (terminal/web).
• [ ] Enhance error handling and logging.
• [ ] Extend toolset (e.g., file handling, browser automation).
• [ ] Multi-tool calling

---

🤝 Credits

• Coagent
• MCP

Enjoy building your intelligent terminal assistant!