capture-learninglisted

# Capture Learning Turns an in-session explanation into a durable artifact in whatever learning-resource location is paired with the *current* project — never assume any specific stack or repo name. Discover everything about the target from its own docs each time. ## 1. Find the paired learning-resource location Check the current project's own memory mechanism, in order, stop at first hit: 1. If a `.beads/` dir exists here: `bd memories learning-resource` / `bd recall learning-resource-location`. 2. Else check this session's file-based memory for a `reference`-type entry pointing at a learning location. 3. Nothing found → ask the user for the path once. Save it back using whichever mechanism the project already uses (`bd remember learning-resource-location "<path> (<what for>)"` if beads, else a `reference`-type memory file) so it's only asked once per project. This same lookup also covers the exercises-root (step 5) — the *location* is stable per project even though *where a given capture fits inside it* is not. ## 2. Read the target location's own conventions Read that location's `CLAUDE.md` (or `README.md` if no `CLAUDE.md`) to learn its actual structure: cheat-sheet file names/paths, style contract, exercise/GOAL.md format, any regen command (e.g. a TOC generator). Do not bring in assumptions from any other stack — every convention comes from what's read here, every session. No `CLAUDE.md`/`README.md` yet (first capture at this location) → ask the user