Skip to content
Back to AI Coding

Sourcegraph Cody for Large Codebases

AI assistant with deep codebase search and context over large monorepos.

AI Coding ToolsIntermediate

Why

Stack Overflow's 2024 survey found 62% of developers currently use AI tools (up from 44% in 2023), with 82% of them using AI to write code. Cody leverages Sourcegraph's code graph and search to fetch precise context across huge repos (millions of LOC) where vector-only tools lose signal.

If your org runs Sourcegraph, Cody gives AI answers grounded in your actual code symbols and references.

When

Use Cody when your codebase is too large for IDE indexing (>500k LOC), when you already run Sourcegraph, or when you need cross-repo 'where is this symbol used' answers. Avoid adopting a new AI tool mid-sprint — evaluate it in a spike first to avoid disrupting in-flight work.

How

1. Install the Cody VS Code/JetBrains extension and auth with your Sourcegraph instance. 2. Chat: ask "where is rateLimit enforced?" — Cody runs a code search, not just embeddings. 3. @repo / @file scope context; Cody fetches relevant snippets via the code graph. 4. Inline edits (Cmd+K) and autocomplete use the same context engine. 5. Custom commands in .cody/commands/*.md define reusable prompts.

Example

.cody/commands/find-owners.md: Ask: "who are the CODEOWNERS of the files touched by this change?" Use the code graph to list owners and recent commit authors. In chat: @repo "summarize the auth flow and list every middleware in the chain"

Common Problems

  • ⚠Error: Cannot find module 'xxx' → run `npm install xxx` and restart the editor/extension host.
  • ⚠Error: "Extension not activated" / greyed-out AI features → check the API key setting and network; run the extension's "Check Connection" command.
  • ⚠Error: Rate limit 429 → upgrade plan or reduce request frequency; check the tool's usage dashboard for quota.

Best Practices

  • ✓Scope with @repo or @file to keep context precise and fast.
  • ✓Put reusable prompts in .cody/commands and version them in git.
  • ✓Measure the outcome with a concrete metric (time saved, defect rate, test coverage) so the practice can be justified and improved.

FAQ

What is Sourcegraph Cody for Large Codebases?

AI assistant with deep codebase search and context over large monorepos. It helps developers quickly grasp the core value and use cases of Sourcegraph Cody for Large Codebases.

How do I get started with Sourcegraph Cody for Large Codebases?

To get started with Sourcegraph Cody for Large Codebases, focus on these points first: Install the Cody VS Code/JetBrains extension and auth with your Sourcegraph instance.

What are common mistakes with Sourcegraph Cody for Large Codebases?

Common mistakes when using Sourcegraph Cody for Large Codebases include: Error: Cannot find module 'xxx' → run `npm install xxx` and restart the editor/extension host.; Error: "Extension not activated" / greyed-out AI features → check the API key setting and network; run the extension's "Check Connection" command.; Error: Rate limit 429 → upgrade plan or reduce request frequency; check the tool's usage dashboard for quota..

What are the best practices for Sourcegraph Cody for Large Codebases?

Best practices for Sourcegraph Cody for Large Codebases include: Scope with @repo or @file to keep context precise and fast.; Put reusable prompts in .cody/commands and version them in git.; Measure the outcome with a concrete metric (time saved, defect rate, test coverage) so the practice can be justified and improved..

How do I learn more about Sourcegraph Cody for Large Codebases?

Start with the official docs, then search real-world examples on GitHub. Validate patterns in a small scratch repo before applying them to production code.