Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Lemda AI 🧠

Nepal's First Autonomous AI Terminal Agent

Lemda AI is a powerful, production-ready coding assistant that lives directly in your terminal. With a beautifully designed Terminal User Interface (TUI), native PC access, and a multi-model gateway, Lemda acts as your AI peer programmer, capable of executing commands, reading files, writing code, and answering complex queries.

🚀 Features

  • Beautiful TUI: Built with Rust (ratatui and crossterm), featuring a live status bar, visible cursor, chunked typing effects, and message queueing.
  • Multi-Model Gateway: Connects to multiple LLMs seamlessly. Supports:
    • Gemini 2.5 Flash (Advanced, FREE)
    • Gemini 2.5 Pro (Ultra, PRO)
    • Gemini 2.0 Flash Lite (Fast, FREE)
    • DeepSeek V4 (Reasoning, FREE)
  • Local PC Access: True autonomous capability. The agent can use tools like run_shell, read_file, and write_file to navigate your computer, check your OS, and write real code to your hard drive.
  • Interactive Command Menu: Type / to open a dynamic popup menu to switch models, change levels, and access more modes on the fly.
  • Auto-Sync Authentication: Simple browser-based login flow. Paste your API key in the browser and watch the terminal automatically sync and boot up.

📁 Monorepo Architecture

Lemda is structured as a modern Rust/TypeScript monorepo:

lemda/
├── apps/
│   ├── docs/          # Documentation website
│   └── ui/            # Web Dashboard (React/Next.js)
├── crates/
│   ├── agent/         # (Legacy) Core agent testing logic
│   ├── backend/       # Rust Axum Proxy - The Multi-Model Gateway
│   └── cli/           # Rust TUI Application - The Terminal Client
└── dont-touch-ai/     # Reserved internal services

🛠️ Quick Start

1. Setup the Backend Proxy

The backend handles routing requests to Google Gemini and DeepSeek securely.

  1. Navigate to the UI folder and create an .env file:
    cd apps/ui
    touch .env
  2. Add your API keys to apps/ui/.env:
    GEMINI_API_KEY=your_gemini_key_here
    DEEPSEEK_API_KEY=your_deepseek_key_here
  3. Start the backend server:
    cd crates/backend
    cargo run

2. Install and Run the CLI

The CLI is your primary interface. Install it globally on your machine:

cd crates/cli
cargo install --path . --force

Now, simply open any terminal window and type:

lemdaai

3. Usage

  • Type natural language prompts and press Enter.
  • Use the Up/Down arrows to navigate chat history (Coming Soon).
  • Type / to open the Command Menu.
  • Select /model to switch between Gemini and DeepSeek.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

About

autonomous AI agent engineered to navigate, execute, and solve complex, end-to-end tasks directly within a terminal environment. By combining dynamic action execution with context-aware tool usage, Lemda achieved #1 on the Terminal-Bench benchmark, outperforming baseline models on complex CLI workflows.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages