Lit

Lit (Let's do IT) is a LLM-powered command-line agent that runs directly in your terminal.

It can read and modify files, execute shell commands, run code, and help you complete programming tasks through an interactive terminal interface.

---

Features

• Base file reading, writing, and editing
• Beautiful TUI with cross-platform terminal support
• Persistent project memory
• Plugin support

---

Quick Start

We strive to make Lit easier to use. Since configuring and using other agent software often involves tedious setup and various limitations, we have focused on providing the best possible out-of-the-box experience.

Prerequisites

• A terminal with truecolor support
• A terminal with truecolor support
• A terminal with truecolor support
• A terminal with truecolor support
• A terminal with truecolor support
• uv

Install

We use uv to manage the Python environment, so you’ll need to install it first:

pip install uv

Then, you need to clone this repo.

# Clone the repository
git clone https://github.com/the-OmegaLabs/lit.git
cd lit

# Install dependencies
uv sync

Configure

Open the Lit folder and edit config.py to set your API key and model:

# API server endpoint URL. Require an OpenAI-compatible endpoint:
base_url = 'https://api.example.com/api/v1'

# API authentication key. Used to verify your identity when calling the API.
api_key = 'your-api-key-here'

# Model displaying name. Specifies which AI model in system prompt and UI.
model = 'gpt-5.6-sol - Example Provider'

# Model identifier id. Specifies which AI model in requests.
model_id = 'gpt-5.6-sol'

# ==========================
# UI Configuration
# ==========================

# Whether to enable the alternative theme.
#
# False:
#   Use the default color theme: Magentic
#
# True:
#   Switch to the alternate theme: Ocean
use_alt_theme = False

You're all set.

Run

uv run -m lit

or open the lit folder and run the app.py directly.

Remote MCP Servers

Lit loads remote MCP tools through plugins/mcp_server.py. Add each server to the root-level mcp.json, then restart Lit. The plugin discovers the remote tools during startup and exposes them to the agent just like built-in plugins.

{
  "mcpServers": {
    "company-tools": {
      "url": "https://mcp.example.com/mcp",
      "transport": "streamable_http",
      "headers": {
        "Authorization": "Bearer ${MCP_API_KEY}"
      },
      "timeout": 30
    }
  }
}

streamable_http is the default transport; legacy sse servers are also supported. Set enabled to false to keep a server in the file without connecting to it. Values written as ${VARIABLE} are read from the process environment, keeping credentials out of mcp.json.

---

🤝 Contributing

Lit is an open-source project led by Omega Labs. Contributions, issues, and feature requests are welcome.

---

📄 License

License information will be announced later.