discover

What it does

/cl-dev:discover populates your knowledge base by analyzing the codebase. It reads through your repository and captures what it finds as durable project knowledge: the project's shape and purpose, the conventions the code follows, and (for a repository with a UI) its components.

The result is a knowledge base the rest of the workflow can lean on. Once it exists, every skill that plans, implements, or reviews work has grounded context about how your project is built, so its output fits your codebase instead of guessing at it.

When to use it

Reach for /cl-dev:discover right after you've set up a repository with /cl-dev:init, to do the bulk initial populate of the knowledge base. It's the fastest way to go from an empty knowledge base to one that reflects your actual codebase.

Come back to it later when you want a full re-scan, for example after a large refactor or a stretch of change that's left the knowledge base out of step with the code.

For knowledge that doesn't live in the code (the reasoning behind a decision, tacit team conventions, or a departing teammate's context), use /cl-dev:extract instead, which draws that out through an interview rather than a scan.

How to invoke it

From your coding agent, in the repository you want to analyze:

/cl-dev:discover

Run it bare. It scans the current repository on its own. A couple of things to have in place first:

• The repository is set up with CodeLantern. Run /cl-dev:init first if you haven't; discover builds on the knowledge base structure that init creates.
• You're pointed at the code you want captured. discover reflects what's in the repository, so run it once the codebase is in a representative state.