kk-bootstrap

You are doing a one-time bootstrap of this project's knowledge base from its existing documentation. The user invoked this skill in their normal session, so they are watching and can correct you in-flight if you go off track.

Your task

Survey the project's existing markdown documentation, extract candidate knowledge nodes, and write them as new node files directly under nodes/. The user reviews each written file and accepts by leaving it in place or rejects by deleting it. You will work judgmentally, exploring, sampling, following cross-references, not exhaustively. This is a one-pass operation, supervised, and you are the LLM doing the extraction — there is no sub-agent and no runner.

Inputs

• An optional path argument from the user. If provided, treat that as the root of the docs scope. If absent, default to the repo root (the finddocs primitive already filters out non-knowledge content via .gitignore, .kkignore, and a static skip list — see step 1).

Configuration

Before you start, read .ai/kenkeep/config.yaml (falling back to ~/.config/kenkeep/config.yaml) for any user preferences (tags vocabulary, scope hints, etc.). Apply what is relevant.

Resolve the project root