tabbyml/tabby

Tabby is a self-hosted AI coding assistant that you run on your own server or workstation, providing an open-source alternative to GitHub Copilot. Instead of sending your code to a third-party cloud service, Tabby processes everything locally, which matters for companies with strict privacy or data security requirements. It provides inline code completion suggestions as you type in your editor, the same core feature as Copilot, but also includes a chat interface for asking questions about code, a code browser connected to your repositories, and an "Answer Engine" that can index your internal documentation and codebase so it can answer engineering questions specific to your organization. It integrates with GitHub and GitLab for context about pull requests and issues, and supports LDAP for company authentication. Tabby is self-contained and does not require an external database or cloud service to run. It exposes an OpenAPI interface (a standardized REST API format) so it can plug into existing development infrastructure. It supports consumer-grade GPUs, including Apple Silicon's M1/M2 chips via Metal acceleration, making it practical to run on developer laptops as well as servers. Someone would use Tabby when they want GitHub Copilot-style AI assistance but need to keep code private, when they want to customize the AI model used, or when they want to build a team-wide coding assistant with access to internal repositories and documentation without sending that code to an external AI service. The server is written in Rust. IDE extensions for VS Code, JetBrains IDEs, and Vim are available. Docker images are also provided for easy deployment.