`meetings/2026-05-08_090021-morning.md`

55 lines

# DW Morning Stand-Up — Fri 2026-05-08 09:00 PDT

---

## 1. Carryover from Last Meeting

⏳ **Wallpaper cluster + ventura root cause** — No fix landed overnight. ventura-corridor hit 101 restarts at 5pm; current pm2 snapshot doesn't include it (outside top 30), so count is unknown this morning.

⏳ **Steve: Doctor apply-templates.js yes/no** — Day 7 blocked. Lawyer Iter 46 still gated behind this decision.

⏳ **Steve: george-gmail OAuth rotation** — Was at 12 restarts at 5pm; current snapshot shows `george-gmail online restarts=0 uptime=49h`. Likely restarted clean — but token is still stale and will crash again.

⏳ **Freddy port rotation** — Soak window expired Thu evening. Script is ready (`rotate-pg-scoped.sh`). No blocker except execution.

---

## 2. Overnight

**Crashes / instability** — Current top-30 snapshot shows all listed processes at restarts=0. No errored agents visible. However, ventura-corridor is absent from the snapshot entirely — its status is unknown. ⚠️ Do not assume it's stable.

**Long-running jobs** — graphic-loop pass 13 landed on 1890s–1980s sites (footer brand, FB link, OG/Twitter meta). Decade sites 1920s/1930s/1970s also received a separate "Scrub Designer Wallcoverings from products.json" pass. NationalPaperHangers got `rel=alternate` JSON + HTTP Link header. novasuede nav collapsed to hamburger across all widths.

**External signals** — No remote commits from collaborators. Three open credential TODOs in `_shared`: PG `dw_admin` password, Gemini key #1, Gemini key #2 — none are rotated yet.

---

## 3. Today's Plan

`debugger-agent` — pull ventura-corridor pm2 logs + shared `server.js` diff vs stable wallpaper sites — accelerating crash pattern needs root cause before weekend — named dependency or code path causing restarts

`Steve` — rotate george-gmail OAuth token (~10 min) — token is stale; next crash resets the 49h uptime clock — process stable after rotation, restarts stay at 0

`Steve` — Doctor apply-templates.js yes/no decision (~10 min) — day 7 block; lawyer Iter 46 can't advance — decision made, agent unblocked

`yolo-runner or Steve` — execute Freddy port rotation via `rotate-pg-scoped.sh` — soak window expired Thu evening, no blocker — Freddy using scoped DB user, no shared-DB risk

`Steve` — mint Gemini key #1 at aistudio.google.com/apikey, paste via `/secrets` — hardcoded key #1 was in 7 Patty routes; unrotated = live credential risk — routed and validated by secrets skill

`graphic-loop` — continue decade site passes (pass 14) — momentum is strong; 1800s–1980s all at pass 13 — commits land on remaining gap sites

---

## 4. Blockers Steve Owns

- **Doctor apply-templates.js** — 7-day decision block; cannot delegate
- **george-gmail OAuth** — token rotation requires Steve's Google account
- **Gemini key #1 + #2** — must mint fresh at aistudio.google.com; can't be scripted without Steve login
- **PG dw_admin password** — runbook ready at `~/Projects/_shared/docs/pg-rotation-runbook.md`; 32-min window; Steve must authorize per-action

---

## 5. One Thing to Think About

ventura-corridor was at 101 restarts at 5pm and is absent from this morning's top-30 snapshot — it may have crashed out of rotation entirely or been restarted and not yet accumulated count. The wallpaper cluster (30+ sites) has had no shared root-cause diagnosis for multiple days. If this is a shared dependency in `server.js`, the weekend is a bad time to let it compound. Is it worth a 30-minute focused agent run this morning to isolate the common failure before Steve steps away?