extract

/cl-dev:extract sits a person down and interviews them for what they know about a project, then writes that knowledge into your CodeLantern knowledge base so the rest of the workflow can draw on it.

What it does

The skill runs a guided interview with a developer and turns their answers into knowledge-base entries, the shared project context that later skills read from. The interview is a two-way conversation: you talk about the work, and the skill asks follow-up questions to draw out the details that usually live only in someone's head.

The interview ranges broadly across the kinds of knowledge a project accumulates: the vocabulary the team uses, the decisions behind why things are the way they are, the conventions people follow, the recipes for common tasks, the components that make up the system, and the plain facts about the project. What comes out gets written across the knowledge base, ready for the workflow skills to lean on.

When to use it

Reach for extract when you have knowledge in a person's head that isn't written down anywhere:

• Seeding a new project. When you're first setting a repository up with CodeLantern and want the knowledge base to start with real context rather than an empty shell.
• Capturing a departing teammate. When someone who knows the project well is moving on, run extract with them to record the tacit knowledge before it walks out the door.

It pairs naturally with /cl-dev:discover, which scans the codebase itself to populate the knowledge base. extract covers the things a scan can't see: the reasoning, the history, and the unwritten conventions. Use both to seed a project well: discover for what the code shows, extract for what only a person knows.

How to invoke it

Run the command inside your coding agent, in the target repository, with the developer being interviewed present to answer:

/cl-dev:extract

A couple of things to have in place first:

• The repository is set up with CodeLantern. If you haven't done this yet, run /cl-dev:init.
• Someone with knowledge to share is at the keyboard. The skill draws knowledge out through conversation, so it needs a person who can answer its questions.