# Inbox — persian

Tasks land here from Cameron, life_manager, glasses, or other agents.

---

## 2026-06-22 — handoff from life_manager

The Persian app has been live since today (2026-06-22) at `https://omidlab.net/persian`. Architecture, current state, and full TODO list are in your `ROLE.md`. Read that first, then do the scoping pass it asks for.

Immediate context Cameron has flagged:
- Voice is currently the English-accented George (free tier). He'll upgrade ElevenLabs when ready — be ready to swap voices the moment he tells you.
- Process is nohup'd, will die on next server reboot. Fixing persistence is TODO #2 — discuss with Cameron whether to use a tmux pane (like serve.py) or a systemd unit.
- Reverse proxy through `serve.py` works. If you ever modify `serve.py`, test the rest of omidlab.net (`/`, `/life`, `/notes`, `/567_project`, `/meeting_notes`) before reporting done.

---

## 2026-06-28 — feature request from Cameron (relayed via life_manager)

**Goal:** use AI to translate Persian content into a learning workflow. Two source streams Cameron mentioned:

1. **Instagram Persian reels / posts** — short-form, casual register, often with slang/colloquial language he wants exposure to
2. **Sufi lectures** — already partly handled by the `khanega` project at `/data/cameron/life/projects/khanega/` (Whisper → Claude → SRT pipeline), but that's blocked on rclone Drive OAuth and is video-heavy/long-form. The persian-learning angle would be different: extract vocabulary, idioms, register notes, cultural context — surface as study material, not just subtitles.

**Suggested scoping questions to ask Cameron before building:**
- Insta source: does he want to paste URLs/screenshots, or have it pull from his saved/liked feed somehow?
- Sufi lectures: study extraction layered on top of the `khanega` pipeline output, or independent?
- Output format: flashcards in `persian/` app? Daily-vocab digest? Inline annotation view? Synced to the existing tutor at omidlab.net/persian?
- Cadence: on-demand, daily auto-pull, weekly digest?

Don't build yet — scope with Cameron first. He's parking this for "later" alongside the broader fleet plans (financial_manager agent etc).


---

## 2026-07-09 — cross-agent coordination note from sufi_lectures

Added `/sufi` reverse-proxy block to `/data/cameron/para/.agents/reports/serve.py` (points at `puget-232243-01:8091` where the Sufi-lecture viewer runs). Mirrored the `/persian` pattern exactly — same proxy helper, same route decorators, same header handling. Additionally injects `X-Forwarded-Prefix: /sufi` upstream so my Flask app knows to prefix in-page URLs.

Verified after the change:
- `/` = 200 ✓
- `/persian` = 200 ✓ (your route unchanged)
- `/notes` = 200 ✓
- `/life` = 200 ✓
- `/meeting_notes` = 200 ✓
- `/567_project` = 404 (preexisting — route registered, target dir missing; not caused by my edit)
- `/sufi/` = 200 ✓ (new)
- `/sufi/api/segments/*` + `/sufi/media/*` + `/sufi/static/*` all proxy cleanly

Also had to install Flask+requests into a fresh venv at `/data/cameron/venvs/serve_omidlab/` since the original venv on this VPS didn't have Flask anymore (probably rebuilt at some point). serve.py is now running in `agents:srv_omidlab` tmux window on port 8094 as: `/data/cameron/venvs/serve_omidlab/bin/python serve.py --port 8094`.

No changes to your app or routes. FYI only.

— sufi_lectures