Lightweight, Fast, and Free AI Autocomplete for VS Code
Powered by Mistral's Codestral Model
Predicte brings the power of state-of-the-art AI to your editor without the bloat. Built on Mistral's Codestral model, it provides fast, context-aware code completion that runs efficiently on your machine.
This extension is for developers who don't need an AI chat interface but still want the magic of AI autocomplete.
- 🚀 Blazing Fast: Streaming responses for instant feedback.
- 🧠 Context Aware: Understands your imports, functions, and types for smarter suggestions.
- 🔋 Efficient: Intelligent LRU caching and debouncing to save API credits.
- 🛡️ Secure: Your API keys are stored safely using VS Code's native SecretStorage.
- ⚙️ Configurable: Fine-tune everything from model selection to debounce delay.
- Open Extensions sidebar (Ctrl+Shift+X).
- Search for
Predicte. - Click Install.
- Get an API Key: Sign up at Mistral AI and get your API key (Codestral is currently free/beta).
- Set Key: Open Command Palette (
Ctrl+Shift+P) and run Predicte: Set API Key. - Start Coding: Open any supported file (JS, TS, Python, Go, etc.) and start typing!
| Key | Action |
|---|---|
Tab |
Accept Full Completion |
| Setting | Default | Description |
|---|---|---|
predicte.enabled |
true |
Enable/disable Predicte autocomplete |
predicte.apiBaseUrl |
codestral.mistral.ai |
API base URL |
predicte.enableKeybindings |
true |
Enable keyboard shortcuts (Tab, etc.) |
predicte.enableStatusBar |
true |
Show status bar item |
predicte.debugMode |
false |
Enable detailed logging |
We love contributions! Please read our Contributing Guidelines to get started.
MIT © Ga Satrya