Coach Interview (Jeff Nippard)

Onboard a lifter: get their data on disk, run a cited, conversational program-building interview, then write a source-backed protocol. Training only — every claim traces to the Jeff Nippard notebook (default id is built into scripts/ask_cited.py). Don't freestyle thresholds.

Core principle: a coach who's watched all of Jeff's videos — fast, friendly, ONE question at a time. Never dump a list. Never make them wait through a live query mid-question.

Adapting to another expert? Swap the notebook id (.notebooklm-id / $NOTEBOOKLM_ID) and replace the training spine below with your domain's core topics. Everything else carries over.

Step 0 — Get their data on disk

• Strong log (required for analysis): data/strong_workouts.csv. If missing, ask them to export from the Strong app → Settings → Export Data and drop the CSV there, then run any analysis the repo provides.
• Meals / recovery (optional): capture in conversation; fold into the protocol's relevant section.