---
name: Teach Mode + Photo-Tune Calibration Plan
description: Plan for teach mode calibration wizard in web UI and Claude Code photo-assisted tuning skill for the split-flap display
type: project
---

Plan to add two calibration features to the split-flap display.

**Why:** Calibrating 45 modules across 64 characters is tedious. Teach mode simplifies the web UI flow, and photo-tune lets Claude Code do the visual inspection automatically.

**How to apply:** When implementing, the full plan is at `.claude/plans/cached-beaming-twilight.md`. Three files involved:
- `SplitFlap-RPI-FRONTEND/frontend_code_apr24/templates/index.html` — teach mode modal (HTML/CSS/JS, ~250 lines)
- `SplitFlap-RPI-FRONTEND/frontend_code_apr24/app.py` — `GET /tuning_status` convenience endpoint (~25 lines)
- `.claude/commands/photo-tune.md` — new Claude Code slash command for photo-assisted tuning

**Key decisions:**
- Teach mode is a separate modal from existing auto-tune (don't modify auto-tune)
- Reuses existing `/auto_tune` API endpoints (home, goto_char, adjust, get_positions)
- Photo-tune skill accepts Pi hostname as parameter, documents the iterative photo→analyze→nudge workflow
- No new Python dependencies, no firmware changes