Back to docs

Context & codebase

Zod uses a custom embedding model to index your codebase, giving agents best-in-class recall across large codebases. Here's how it works.

Codebase indexing

When you open a project in Zod, it automatically indexes your codebase. This process creates a searchable index that agents use to understand your project structure, dependencies, and conventions.

Automatic

Indexing happens in the background as you work. No manual setup required.

Incremental

Only changed files are re-indexed. Fast updates even in large codebases.

Secure

Indexing happens locally. With Privacy Mode, your code is never sent externally.

Multi-language

Supports JavaScript, Python, TypeScript, Go, Rust, Java, and 20+ languages.

Semantic search

Ask questions in natural language and Zod finds the relevant code, documentation, and context. Unlike regex or keyword search, semantic search understands code semantics.

  • Find functions by what they do, not just their name
  • Discover related code across the entire codebase
  • Understand type relationships and dependencies
Try asking Zod:
"Where do we handle user authentication?"
Found in auth source files including login, session, and middleware modules.
"Find all places where we format dates"
7 matches across utils, components, and API layer

Privacy mode

When privacy mode is enabled, your code is never used for training. Zod maintains zero data retention agreements with all model providers.

Privacy Mode ON

No code data is sent to model providers for training. Zero data retention. Your codebase is indexed locally and only relevant context is sent with each request.

Privacy Mode OFF

Your code may be used to improve models. You can toggle Privacy Mode at any time in Settings.