How we index your codebase
When you first open a project with Axon, it does something most AI coding tools don't: it reads your entire codebase.
Not just the files you have open. Not just recently edited files. Everything.
This is called codebase indexing, and it's the foundation of everything Axon does well.
What indexing actually does
Indexing creates a semantic map of your codebase. Axon parses every file, extracts the structure — functions, types, exports, imports, call graphs — and builds a representation that it can query in real time as you type.
This is different from just storing files. Axon understands relationships. It knows that getUserProfile calls fetchUser, which depends on validateSession. When you're writing code that touches the auth system, Axon can surface relevant context from across your entire codebase in milliseconds.
Why it's fast
The first index takes a few seconds on most codebases. After that, Axon maintains an incremental index — only re-indexing files that have changed. This means the index is always up to date without any noticeable performance impact.
We've optimized the indexing pipeline significantly over the past year. What used to take 30 seconds on a 100k-line codebase now takes under 5.
Privacy
Your code never leaves your machine during indexing. The index is built and stored locally. When Axon needs to make a suggestion, it sends only the relevant context to the model — never your entire codebase.
This is a hard constraint we've built around from the beginning. Your code is yours.