Cody — AI Coding Assistant by Sourcegraph
Sourcegraph Cody is an AI coding assistant combining LLM power with deep codebase understanding via code search. For VS Code, JetBrains, and web.
✅ Pros
- + Deep codebase context via Sourcegraph's code intelligence graph
- + Handles large monorepos better than most AI tools
- + Supports Claude 3.5 Sonnet, GPT-4o, and Gemini
- + Multi-repo context: reference code across multiple repos at once
- + Free tier is genuinely useful (not just a trial)
- + Strong enterprise features: SSO, audit logs, policy controls
⚠️ Cons
- − Best features require Sourcegraph account/integration
- − Pro plan needed for unlimited usage
- − Heavier setup than simple tools like Copilot
- − Context quality depends on Sourcegraph indexing your repo
What is Cody?
Cody is Sourcegraph’s AI coding assistant that combines LLM capabilities with Sourcegraph’s powerful code search and intelligence graph. The key differentiator: Cody can reference your entire codebase — not just the open files — when generating code or answering questions.
Key Features
- Codebase-aware autocomplete: Suggestions informed by patterns across your whole repo
- Chat with context: Ask questions and Cody fetches relevant code from anywhere in your codebase
- Commands: Built-in
/edit,/explain,/test,/smell(find code smells) - Multi-repo context: Enterprise users can query across multiple repos simultaneously
- Model choice: Select Claude 3.5 Sonnet, GPT-4o, or Gemini depending on the task
Free vs Pro
| Feature | Free | Pro ($9/mo) |
|---|---|---|
| Autocomplete requests | Limited | Unlimited |
| Chat messages | 200/month | Unlimited |
| Context window | Standard | Extended |
| Model selection | Limited | All models |
Best For
- Developers working in large monorepos or multi-repo organizations
- Teams already using Sourcegraph for code search
- Enterprise engineering orgs that need audit/compliance controls
- Anyone wanting strong cross-file context at a lower cost than Cursor
Quick Start
Install the Cody extension in VS Code or JetBrains, then sign in with a free Sourcegraph account:
VS Code: Extensions → search "Cody AI" → Install → Sign in
JetBrains: Settings → Plugins → search "Cody" → InstallOnce installed, use the Cody sidebar to chat or trigger commands directly from the editor. The /explain command on any selected block is a fast way to understand unfamiliar code in a large codebase.
# Useful built-in commands (trigger from command palette or sidebar):
/explain → Explain what selected code does
/edit → Rewrite selected code per your instruction
/test → Generate unit tests for the selected function
/smell → Identify potential code quality issues
/doc → Generate docstring/JSDoc for the selectionUse Cases
Navigating large monorepos: In repos with hundreds of services, Cody uses Sourcegraph’s code graph to find the actual implementation of a function across files — not just open tabs. Ask “where is calculateTax defined and what calls it?” and get an accurate answer with links.
Cross-repo context in enterprise: Enterprise teams can configure multi-repo context so Cody understands how a shared library in one repo is consumed by services in other repos. This is especially useful for SDK teams or platform engineering.
Onboarding new developers: New team members can ask Cody natural-language questions about the codebase (“how does our auth middleware work?”, “what’s the standard way to add a new API endpoint here?”) and get context-aware answers grounded in your actual code — not generic documentation.
Code review acceleration: Use Cody to generate explanations of PRs, suggest test cases for changed code, or flag potential issues in a diff — all within the IDE without leaving the review workflow.