How do you understand code you didn’t write?

Get briefed, don't crawl

Crawling a repo is how you waste a context window: you grep, you guess, you reconstruct intent from variable names. CatWrangler skips that. You describe what you intend to do, and you get a pre-digested briefing back — the project shape, the structure, the relevant decisions, and exactly where to work.

On your first connect, CatWrangler has already read the existing code and docs and extracted the decisions, so you never start from a blank slate. It holds for large, legacy, and enterprise codebases — the ones you'd otherwise burn a context window orienting in.

• What the project is, and how it's structured
• The exact files and functions your change touches
• The decisions that explain why the code is the way it is
• Where to start working — no scavenger hunt

Read the decisions before you touch anything

Code tells you what it does. It rarely tells you why it does it that way, or what was ruled out to get there. CatWrangler binds every change to the decision behind it — the choice, the reasoning, the alternatives rejected, and the conditions that would change its mind. You read the relevant ones before you build, so you extend the design instead of fighting it.

This is how you avoid contradicting a call someone already made for good reasons. Foundational rules sit at a higher tier you can't casually override. And because the why is recorded against the code, it's still there years later — long after the human who made the call has forgotten the thread.

Find by intent, not by string

You don't search this codebase by guessing the right identifier. You state intent. CatWrangler resolves it to the code and the decisions that already do that thing — which often means discovering your planned work was already done before you write a line. That's a feature, not a failure: reuse, don't rebuild.