Zero-Friction, Privacy-Aware AI Commit Messages for VS Code
Powered by Mistral AI or Local LLMs (Ollama, vLLM, etc.)
Predicte Commit analyzes your staged changes and generates concise, descriptive commit messages using AI. It brings the magic of AI automation to your git workflow while keeping you in control.
This extension is for developers who want to automate the tedious task of writing commit messages without leaving their source control view.
✨ Features
💸 100% Free: Use Mistral's latest coding models (Devstral 2).
🚀 Zero-Friction: Generate messages with a single click or command.
🔒 Privacy-Aware: Built-in content filtering and explicit file ignores to keep secrets safe.
🤖 Flexible Providers: Use Mistral AI (cloud) or your own Local instance (Ollama, vLLM, LM Studio).
🧠 Smart Truncation: Automatically handles large diffs to stay within context windows.
⚡ Efficient: Streamlined to provide quick results directly in your commit input box.
📦 Installation
Open the Extensions sidebar in VS Code (Ctrl+Shift+X).
Search for Predicte Commit.
Click Install.
🚀 Getting Started
Get an API Key: Sign up at Mistral AI and get your API key.
Set Key: Open Command Palette (Ctrl+Shift+P) and run Predicte Commit: Set API Key.
Stage Changes: Stage your files in the Source Control view as usual.
Generate Message:
Click the Sparkle Icon $(sparkle) in the Source Control title bar.
Or open the Command Palette (Ctrl+Shift+P) and run Predicte Commit: Generate Message.
Review: The generated message will appear in your commit input box. Edit if needed, then commit!
Note: If using the Mistral provider (default), you will be prompted to enter your API key on the first run.
⚙️ Configuration
Setting
Default
Description
predicteCommit.provider
mistral
Choose provider (default: mistral).
predicteCommit.models
[]
List of models to use. If empty, uses provider defaults.
predicteCommit.ignoredFiles
["*-lock.json", "*.svg", "dist/**"]
Glob patterns to exclude from analysis.
predicteCommit.useLocal
false
Enable to use a local provider instead of Mistral.
predicteCommit.localProvider
ollama
Local provider to use when Use Local is enabled.
predicteCommit.localBaseUrl
(empty)
Base URL for local provider (e.g., http://localhost:11434/v1).
predicteCommit.localModel
(empty)
Model name for local provider.
predicteCommit.debugLogging
false
Log prompt payload and diagnostics to the output channel.
