Sourcegraph Cody
Cody is Sourcegraph's AI coding assistant that combines large-scale code search over an entire codebase (including large monorepos) with LLM-based chat, autocomplete, and code editing, using Sourcegraph's search index to ground answers in…
Definition
Cody is Sourcegraph's AI coding assistant that combines large-scale code search over an entire codebase (including large monorepos) with LLM-based chat, autocomplete, and code editing, using Sourcegraph's search index to ground answers in the actual code rather than relying only on the model's training data.
Overview
Cody builds on Sourcegraph's core product — a universal code search engine originally built for navigating massive codebases and monorepos across many repositories — and layers AI chat, autocomplete, and command features on top of that search infrastructure. Its central differentiator is context retrieval at scale: rather than relying only on files open in an editor or a small local embeddings index, Cody can pull relevant context from Sourcegraph's precise code search and code intelligence (symbol definitions, references, code navigation graphs) across an organization's entire code estate, including many repositories at once. This matters most for large organizations with codebases too big to fit in any practical context window and where cross-repository awareness is valuable — for example, understanding how a shared internal library is used across dozens of services before proposing a change. Cody supports natural-language chat about code, inline autocomplete, and 'commands' for common tasks (explain code, generate docstrings, generate unit tests, fix a bug), and it can be configured to work with several LLM providers. Cody is offered both as part of Sourcegraph's enterprise code search platform and as a more accessible product for individual developers and smaller teams, and it competes with other AI coding assistants primarily on the strength of Sourcegraph's code search and context quality — positioning itself for large, complex, multi-repository engineering organizations where 'the AI actually knows the whole codebase' is the key value proposition, rather than general-purpose per-file coding assistance.
Key Features
- Grounded in Sourcegraph's large-scale, cross-repository code search engine
- Context retrieval spans an entire organization's codebase, not just open files
- Chat, autocomplete, and pre-built commands (explain, document, test, fix)
- Code intelligence integration: symbol definitions, references, navigation graphs
- Works across multiple repositories and monorepos simultaneously
- Configurable to use several different LLM providers
- Available as IDE extensions (VS Code, JetBrains) and via Sourcegraph's web UI
- Enterprise deployment options including self-hosted Sourcegraph instances
Use Cases
Alternatives
History
Cody is an AI coding assistant built by Sourcegraph, a company specializing in code search and code intelligence. Cody was first unveiled in June 2023 and reached general availability (version 1.0) on December 14, 2023. Its distinguishing idea is codebase awareness: rather than treating source code as plain text, Cody leverages Sourcegraph's code-graph and search infrastructure to provide context-aware completion, chat, and commands grounded in an entire repository. At general availability it supported models including Anthropic's Claude and OpenAI's GPT-4 Turbo, and it shipped as extensions for editors such as VS Code and JetBrains, positioning itself against tools like GitHub Copilot.
Sources
- Sourcegraph — "Cody is generally available" · as of 2026-07-17
- Sourcegraph — Cody documentation · as of 2026-07-17