Cody by Sourcegraph is built on a fundamentally different premise from other AI coding tools: instead of working with just the file you have open, Cody reads your entire codebase to provide contextually accurate suggestions, explanations, and code generation.
Powered by Claude and other frontier LLMs, Cody understands your project's architecture, internal APIs, naming conventions, and existing patterns. When you ask it to write a new feature, it generates code that actually fits your codebase — not generic boilerplate.
Cody integrates natively with VS Code and all JetBrains IDEs (IntelliJ, PyCharm, GoLand, WebStorm), and also works inside Sourcegraph's own web interface. It offers inline code completions, natural language chat for coding questions, a Commands library for common tasks like writing unit tests and generating docstrings, and Custom Commands you can define for your team's unique workflows.
For enterprises, Cody connects to private GitHub, GitLab, and Bitbucket repositories via Sourcegraph's code intelligence platform, ensuring the AI always has access to the latest version of your codebase without any data leaving your infrastructure.
Cody is the coding assistant for teams who need AI that works within the complexity of real-world software — not just toy examples.
Key Features
Entire codebase context — not just the open file
Powered by Claude, GPT-4, Gemini, and other frontier LLMs (user-selectable)
Inline autocomplete for fast code suggestions
Natural language chat for explaining and writing code
Built-in commands: generate unit tests, write docstrings, explain code