Cody is Sourcegraph's AI coding assistant that actually understands your entire codebase - not just the file you're looking at. Unlike generic code AI tools that guess based on a few lines of context, Cody reads your whole repository to give answers, write code, and explain things with genuine awareness of your architecture. It's like having a senior engineer who joined the team six months ago and already read every line of code, except it takes about three seconds.
Main Features
- Whole-codebase context: Cody indexes your entire repository and uses it as context for every answer - no more AI suggestions that ignore your existing patterns and conventions.
- Natural language code search: Ask questions like 'how does authentication work in this codebase?' and get answers backed by specific file references and code snippets.
- Autocomplete that knows your code: Single-line and multi-line autocomplete that respects your project's variable names, utility functions, and coding style.
- Inline chat and edit: Highlight code, ask Cody to refactor, explain, or fix it, and apply changes directly in your editor with a diff preview.
- Unit test generation: Generate meaningful unit tests that understand your code's logic, edge cases, and existing test patterns - not just boilerplate.
- Multi-model support: Choose from Claude, GPT-4, Gemini, and open-source models - swap models per task based on speed, cost, or capability needs.
- VS Code, JetBrains and Neovim: Works across major editors with consistent experience - pick your tool, Cody follows.
- Enterprise code graph integration: Combines with Sourcegraph's code search platform for org-wide code intelligence, batch changes, and codebase analytics.
Who Should Use It?
- Software engineers: Onboarding to large codebases who need an AI that actually understands the full project, not just a file.
- Senior developers: Who want an AI that respects existing patterns, naming conventions, and architectural decisions.
- Engineering teams at enterprises: With massive monorepos where context-aware AI is the difference between useful and useless suggestions.
- Open-source contributors: Jumping into unfamiliar repositories who need codebase-aware explanations and navigation.
- Tech leads: Conducting code reviews who want AI assistance that understands project conventions and can spot inconsistencies.
- Startup CTOs and solo developers: Shipping fast who need an AI pair programmer that learns their codebase instantly.
