Introduction
GenerativeIDE (GIDE) is a revolutionary AI-powered code editor that runs completely offline on your machine. Built on the powerful foundation of VS Code, GIDE brings cutting-edge AI capabilities directly to your development workflow without requiring an internet connection.
Unlike cloud-based solutions, GIDE ensures your code never leaves your machine. All AI processing happens locally, giving you the privacy and security you need for sensitive projects. Whether you're working on proprietary software, handling confidential data, or simply prefer working offline, GIDE has you covered.
Deep Codebase Understanding
GIDE goes beyond simple text search. Our Hybrid JEPA (Joint Embedding Predictive Architecture) integration provides semantic understanding of your entire codebase.
| Feature | Traditional Approach | GIDE's Approach |
|---|---|---|
| Code Discovery | Regex/grep pattern matching | Semantic vector embeddings |
| Dependency Resolution | Static import parsing | Dynamic call graph analysis |
| Symbol Tracking | Exact string matching | Contextual intent recognition |
| Cross-file Context | Limited to open files | Full project-wide awareness |
| Scalability | Linear search complexity | Indexed semantic retrieval |
In large codebases, traditional search returns hundreds of results. GIDE's semantic understanding identifies only the files relevant to your context enabling faster development and confident code changes.
Installation
GIDE is available for Windows, macOS (Intel & Apple Silicon), and Linux. Download the appropriate version for your operating system:
Quick Start
Get up and running with GIDE in just a few minutes:
- 1
Download & Install
Download GIDE from our website and run the installer for your operating system.
- 2
Open a Project
Launch GIDE and open a folder containing your code. Use
File → Open Folder. - 3
Start the AI Chat
Press
Ctrl/Cmd + Shift + Ito open the AI Chat. - 4
Ask Questions & Generate Code
Start chatting with the AI, Ask questions about your code, request refactoring, or generate new functionality.
System Requirements
GIDE runs locally with AI models optimized for consumer hardware. Here are the recommended specifications:
| Component | Minimum | Recommended |
|---|---|---|
| RAM | 8 GB | 16 GB+ |
| CPU | 4 cores | 8+ cores (M1/M2/M3 or modern Intel/AMD) |
| GPU | Optional | Metal, CUDA, or Vulkan supported GPU |
Ask Mode
Ask Mode lets you have a conversation with your entire codebase. Simply open the AI Chat and start asking questions about your project.
With Ask Mode, you can:
- Understand complex code and how different parts connect
- Get explanations for unfamiliar code patterns
- Ask technical questions about your project architecture
- Debug issues by describing the problem in natural language
- Learn how specific functions or modules work
Ctrl/Cmd + Shift + I — Open AI Chat
Agent Mode
Agent Mode enables the AI to autonomously make changes to your codebase based on your instructions. Describe what you want, and GIDE will analyze, plan, and execute the changes.
With File Mentions
Reference specific files using the @file syntax for targeted changes.
Example: "Refactor @utils.ts to use async/await"
Without File Mentions
Let the AI automatically determine which files need to be modified based on your request.
Example: "Add error handling to all API calls"
Agent Mode will:
- Analyze your codebase to understand the context
- Plan the necessary changes across files
- Execute the edits with real-time progress updates
- Present the changes for your review before applying
Ask Selection
Ask Selection allows you to quickly get AI assistance for any selected code. Simply highlight the code you want help with and use the keyboard shortcut.
Use Ask Selection to:
- Explain what a selected piece of code does
- Refactor or improve selected code
- Fix bugs in the highlighted code
- Convert code to a different pattern or style
- Add comments or documentation
Ctrl/Cmd + L — Ask about the selected code
Settings
GIDE uses VS Code's powerful settings system. All your VS Code settings, preferences, and configurations work seamlessly in GIDE.
Access settings via Ctrl/Cmd + , or through the gear icon in the sidebar.
If you're already a VS Code user, your familiar workflow and preferences carry over automatically.
Keybindings
GIDE uses VS Code's keybinding system. Customize your shortcuts via Ctrl/Cmd + K, Ctrl/Cmd + S.
Ctrl/Cmd + Shift + I — Open AI Chat
Ctrl/Cmd + L — Ask Selection
Themes
GIDE supports all VS Code themes. Browse and install themes from the Extensions marketplace or use the built-in theme picker.
Access themes via Ctrl/Cmd + K, Ctrl/Cmd + T.
Extensions
GIDE is fully compatible with the Open VSX extension marketplace. Install your favorite language support, linters, formatters, and productivity tools.
Access the Extensions view via the sidebar or press Ctrl/Cmd + Shift + X.
VS Code Fork
VS Code OSS
Open-source code editor originally developed by Microsoft
GIDE is built on top of the VS Code open-source repository. We've forked and extended the open-source codebase to integrate AI capabilities while preserving the powerful editing experience developers love.
By building on VS Code OSS, GIDE inherits:
- •Code editing features (IntelliSense, debugging, Git integration)
- •Extensive language support for virtually every programming language
- •Compatibility with Open VSX extensions
- •Familiar interface for developers worldwide
📜 License & Attribution
The VS Code open-source project is originally developed by Microsoft and is licensed under the MIT License.
GenerativeIDE is based on the open-source VS Code (OSS) repository and is not distributed as Microsoft's Visual Studio Code product.
Disclaimer: Microsoft is not affiliated with, nor does it endorse, GenerativeIDE. VS Code is a trademark of Microsoft Corporation.
llama.cpp
llama.cpp
Georgi Gerganov & Contributors
GIDE's local AI inference is powered by llama.cpp, a high-performance C/C++ implementation for running large language models on consumer hardware.
llama.cpp enables GIDE to:
- •Run AI models 100% locally without cloud dependencies
- •Leverage GPU acceleration (Metal on macOS, CUDA on NVIDIA, Vulkan)
- •Achieve fast inference speeds on consumer hardware
- •Support quantized GGUF models for reduced memory usage
llama.cpp is licensed under the MIT License.
Qwen 2.5 Coder
Qwen 2.5 Coder
Developed by Alibaba Cloud (Qwen Team)
GIDE supports Qwen 2.5 Coder as a selectable model for AI-powered code assistance. Qwen 2.5 Coder is a state-of-the-art code-focused large language model developed and maintained by Alibaba Cloud's Qwen Team, specifically designed for programming tasks and code understanding.
Key capabilities of Qwen 2.5 Coder include:
- •Code generation and completion capabilities
- •Support for 90+ programming languages
- •Code reasoning and debugging assistance
📜 License & Attribution
Qwen 2.5 Coder is developed by Alibaba Cloud (Qwen Team) and is licensed under the Apache License, Version 2.0.
You may obtain a copy of the Apache License at: apache.org/licenses/LICENSE-2.0
Disclaimer: Alibaba Cloud and the Qwen project are not affiliated with, nor do they endorse, GenerativeIDE. Qwen 2.5 Coder is an independently developed open-source model that GIDE integrates as one of its supported AI models.