Show the roster. Present the two groups below (one line each). The protocol layer is discipline for any quantitative work; the research skills are heavier machinery for specific tasks.

Ask what to activate. Offer: all twelve, the protocol layer only, the research skills only, an individual selection, or none. Use the host's question/multi-select interface if it has one (in Claude Code, grouped multi-select questions fit within the four-option limit); otherwise ask in plain text. "None" is a fine answer — the library stays available to read or load ad hoc either way.

Ask the scope. Project-level ( in the user's project — active only there) or user-level ( — active in every project). Default to project-level if the user has no preference.

Install. Copy each chosen skill's directory from in this repository into the chosen location, keeping the directory name. Nothing else: no edits to the files, no extra registration step.

Confirm. List what is now active and where. Note that an activated skill still only engages when its work actually comes up; activation makes it discoverable, not mandatory. To deactivate one later, delete its directory from the install location. Re-run this setup any time to add or change the selection.