redai

A terminal-based AI-powered code editor for developers, written in Rust.

redai combines a fast, modern TUI code editor with context-aware code completion and editing powered by large language models (LLMs) such as OpenRouter/CodeStral. It is designed for productivity, code navigation, and seamless AI-assisted coding directly from your terminal.

Features

• Terminal UI: Fast, responsive code editor with file tree navigation using [ratatui].
• AI Code Completion & Editing: Context-aware suggestions and code edits using LLMs (OpenRouter, CodeStral, etc.).
• File Tree: Browse and open files in your project directory.
• Keyboard & Mouse Support: Navigate, edit, and resize panes with keyboard and mouse.
• Visual Diff: Visualize code changes and AI-suggested edits.
• Configurable: Set your LLM provider, model, and API key via environment variables.

Installation

From releases

Linux:

curl -L https://github.com/vipmax/redai/releases/latest/download/redai-linux.tar.gz | tar -xz --overwrite
sudo mv redai /usr/local/bin/
sudo chmod +x /usr/local/bin/redai
redai

Macos:

curl -L https://github.com/vipmax/redai/releases/latest/download/red-universal-apple-darwin.tar.gz | tar -xz --overwrite
sudo mv redai /usr/local/bin/
sudo chmod +x /usr/local/bin/redai
redai

From sources

1. Clone the repository:

git clone https://github.com/vipmax/redai.git
cd redai

2. Install Rust (if not already):

# mac and linux:
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

# windows:
https://rust-lang.org/tools/install/

3. Build the project:

cargo build --release

# move to path
sudo mv target/release/redai /usr/local/bin/
sudo chmod +x /usr/local/bin/redai

redai

4. Set up environment variables:

Create a .env file or export variables in your shell:

OPENROUTER_API_KEY=your_openrouter_api_key
OPENROUTER_BASE_URL=https://openrouter.ai/api/v1   # (optional, default shown)
OPENROUTER_MODEL=mistralai/codestral-2508     # (optional, default shown)

Usage

redai 

Key Features

• File Tree Navigation:
  • Use arrow keys or mouse to navigate files and folders.
  • Press Enter to open a file.
  • q to quit file tree, Esc to deselect.
• Editor:
  • Edit code as in a normal editor.
  • Ctrl+S to save.
  • Ctrl+Space to trigger AI autocomplete/code edit at the cursor.
  • Esc to undo marks or quit.
  • Control + c - copy
  • Control + v - paste
  • Control + f - file search
  • Control + g - global search
• AI Code Editing:
  • The editor sends code context and recent edits to the LLM.
  • The LLM responds with precise code changes, which are applied and visualized.

AI Completion: How It Works

When you trigger AI completion (with Ctrl+Space), redai:

1. Gathers Context:
   • Captures a small and large context window around your cursor.
   • Includes a summary of your recent edits for better continuity.
2. Sends to LLM:
   • Packages the context and recent activity into a prompt.
   • Sends this to the configured LLM (e.g., OpenRouter/CodeStral).
3. LLM Responds:
   • The LLM returns a response in a strict format using special tokens:
     • <|SEARCH|>: The code to find (with <|cursor|> marking your cursor).
     • <|DIVIDE|>: Separates the search and replacement.
     • <|REPLACE|>: The replacement code.
   • Example:

     <|SEARCH|>let <|cursor|> = 10;<|DIVIDE|>let x = 10;<|REPLACE|>
     

4. Applies the Edit:
   • redai parses the LLM's response, computes the minimal edit, and applies it to your code.
   • The change is visually highlighted in the editor for review.

This approach ensures that completions are context-aware, minimal, and easy to review, making AI assistance both powerful and safe.

Configuration

redai is configured via environment variables:

• OPENROUTER_API_KEY (required): Your OpenRouter API key.
• OPENROUTER_BASE_URL (optional): LLM API endpoint. Default: https://openrouter.ai/api/v1
• OPENROUTER_MODEL (optional): LLM model name. Default: mistralai/devstral-medium-2507

You can use a .env file in the project root for convenience.

Dependencies

• ratatui - Terminal UI
• ratatui-code-editor - Code editor widget
• async-openai - OpenAI-compatible LLM client
• crossterm - Terminal input/output
• tokio - Async runtime
• similar - Diffing
• [serde, serde_json] - Serialization
• tui-tree-widget - File tree navigation

Acknowledgments

• Inspired by modern code editors and the power of LLMs for code assistance.
• Uses OpenRouter for LLM backend.

---

MIT License. Contributions welcome!