Sourcegraph Cody

Unlike AI coding tools that operate only within the currently open file, Cody understands the broader project context including relationships between components, existing API patterns, shared utilities, and architectural conventions to produce suggestions that are genuinely consistent with the codebase's existing style and structure.

This deep codebase understanding is made possible by Sourcegraph's underlying code search and indexing technology.

Cody provides multi-line and single-line code autocompletion as developers type in supported IDEs, significantly accelerating the coding process by predicting not just the next line but entire function implementations, test cases, and boilerplate patterns.

The chat interface allows developers to ask natural language questions about unfamiliar code, request explanations of complex functions, get documentation written for existing methods, or ask Cody to generate new code based on a description of intended behavior.

Commands enable common tasks like writing unit tests, generating docstrings, or refactoring selected code blocks to be executed with a single action.

The platform integrates with Visual Studio Code and the JetBrains family of IDEs as extensions, as well as being accessible through the Sourcegraph web application. This multi-surface availability means developers can use Cody whether they prefer local IDE development or cloud-based code browsing and review.

The context window capabilities extend to handle large codebases, providing up to 500,000 tokens of context for the Free plan and up to 1 million tokens for Cody Pro subscribers enabling meaningful assistance even on complex, sprawling enterprise projects.