Refact Agent is a free, open-source AI Agent that handles engineering tasks end-to-end. It deeply understands your codebases and integrates with your tools, databases, and browsers to automate complex, multi-step tasks.
Refact Agent works effortlessly with the tools and databases you already use:
- 📁 Version Control: GitHub, GitLab
- 🗄️ Databases: PostgreSQL, MySQL
- 🛠️ Debugging: Pdb
- 🐳 Containerization: Docker
- ✅ Deploy On-Premise: Maintain 100% control over your codebase.
- 🧠 Access State-of-the-Art Models: Supports Claude 3.5 Sonnet, GPT-4o, o3-mini, and more.
- 🔑 Bring Your Own Key (BYOK): Use your own API keys for external LLMs.
- 💬 Integrated IDE Chat: Stay in your workflow, no need to switch between tools!
- ⚡ Free, Unlimited, Context-Aware Auto-Completion: Code faster with smart AI suggestions.
- 🛠️ Supports 25+ Programming Languages: Python, JavaScript, Java, Rust, TypeScript, PHP, C++, C#, Go, and many more!
📜 View Full List of Supported Models
📢 Using AI for work? Let’s bring it to your company!
Fill out this form — Our AI Agent will be tailored to your company’s data, learning from feedback, and helping organize knowledge for better collaboration with your team.
- 🚀 Core Features and Functionality
- 🤖 Which Tasks Can Refact Help You With?
- ⚙️ QuickStart
- 🐳 Running Refact Self-Hosted in a Docker Container
- 🔌 Getting Started with Plugins
- 📖 Documentation
- 🥇 Contribution
- 🎉 Join the Community
✅ Unlimited accurate auto-completion with context awareness – Powered by Qwen2.5-Coder-1.5B, utilizing Retrieval-Augmented Generation (RAG).
✅ Integrated in-IDE Chat – AI deeply understands your code and provides relevant, intelligent answers.
✅ Integrated with Tools – Works with GitHub, GitLab, PostgreSQL, MySQL, Pdb, Docker, and shell commands.
✅ State-of-the-Art Models – Supports Claude 3.5 Sonnet, GPT-4o, o3-mini, and more.
✅ Bring Your Own Key (BYOK) – Use your own API keys for external LLMs.
-
🏗 Generate code from natural language prompts (even with typos).
-
🔄 Refactor code for better quality and readability.
-
📖 Explain code to quickly understand unfamiliar code.
-
🐞 Debug code to detect and fix errors faster.
-
🧪 Generate unit tests for reliable code.
-
📌 Code Review with AI-assisted suggestions.
-
📜 Create Documentation to keep knowledge up to date.
-
🏷 Generate Docstrings for structured documentation.
You can install the Refact repository without Docker:
pip install .
For GPU with CUDA capability >= 8.0 and flash-attention v2 support:
FLASH_ATTENTION_FORCE_BUILD=TRUE MAX_JOBS=4 INSTALL_OPTIONAL=TRUE pip install .
The easiest way to run the self-hosted server is using a pre-built Docker image.
See CONTRIBUTING.md
for installation without a Docker container.
- Download Refact for VS Code or JetBrains.
- Set up a custom inference URL:
http://127.0.0.1:8008
- Configure the plugin settings:
- JetBrains: Settings > Tools > Refact.ai > Advanced > Inference URL
- VSCode: Extensions > Refact.ai Assistant > Settings > Address URL
For detailed guidance and best practices, check out our documentation.
Want to contribute to our project? We're always open to new ideas and features!
- Check out GitHub Issues – See what we're working on or suggest your own ideas.
- Read our Contributing Guide – Check out
Contributing.md
to get started.
Your contributions help shape the future of Refact Agent! 🚀
We're all about open-source and empowering developers with AI tools. Our vision is to build the future of programming. Join us and be part of the journey!
📢 Join our Discord server – A community-run space for discussion, questions, and feedback.
Made with ❤️ by developers who automate the boring, so you can focus on building the future.