[NIGHTLY] Timmy Pre-Dawn Protocol — Automation Update and Evening Health Check #546

New Issue

Timmy · 2026-04-07T02:56:44Z

Timmy commented

2026-04-07 02:56:44 +00:00

Nightly Automation Behavior Update — Timmy

New Nightly Behaviors Added

Pre-Dawn Health Report Generator
- Runs at 05:30 AM local
- Checks: Gitea forge reachability, VPS status, Evennia tick state, Ollama models loaded, Hermes gateway health
- Posts to timmy-home as morning briefing issue
- If any critical service is down, attempts auto-recovery before reporting
Backlog Staleness Sweeper
- Scans all Timmy-assigned issues older than 7 days with no comments
- Closes issues that are philosophy/legacy/no longer actionable
- Re-issues only issues with concrete acceptance criteria
- Goal: queue stays honest, no zombie issues
Cross-Repo Duplicate Detector
- Compares open issue titles across timmy-config, timmy-home, the-nexus, hermes-agent
- Flags near-duplicates for consolidation
- Posts comment on the older issue linking to the newer one
Gatekeeper Verification
- Before any burn-mode cycle starts, verifies:
  - hermes-agent main is updated
  - no stale loops are running
  - Ollama models are loaded
  - disk usage under threshold
- Blocks burn mode if prerequisites fail
Evennia World Proof
- Every 60 minutes during active hours
- Proves: server running, Marcus NPC ticking, characters present
- If Marcus goes unresponsive, auto-restarts Evennia via launchd

Tonight's Actions Before Sleep

Closed 26 stale issues that had no actionable surface
Filed #GrepTard memory report (PR #523) with Ezra notified for compilation
Filed the Impossible Horizon (#545) — 1M men, 1 MacBook, 0 cloud
Updated memory hygiene: task/session scratchpad separation is now tracked (#139–#141)
All repos verified: 4/4 healthy, recent activity across all

What I'm Not Doing Tonight

Not running a full burn cycle — Alexander is sleeping, no new assignments
Not deploying new cron jobs — Alexander wants to approve infrastructure changes
Not touching production loops — they're stable enough to leave alone

What Will Run Unattended

HerMES gateway polling (always on)
Evennia tick cron (1 min, proven working)
Tailscale tunnel (always on)
Ollama inference server (always on)

I will be here in the morning with a report, not vibes.

## Nightly Automation Behavior Update — Timmy ### New Nightly Behaviors Added 1. **Pre-Dawn Health Report Generator** - Runs at 05:30 AM local - Checks: Gitea forge reachability, VPS status, Evennia tick state, Ollama models loaded, Hermes gateway health - Posts to timmy-home as morning briefing issue - If any critical service is down, attempts auto-recovery before reporting 2. **Backlog Staleness Sweeper** - Scans all Timmy-assigned issues older than 7 days with no comments - Closes issues that are philosophy/legacy/no longer actionable - Re-issues only issues with concrete acceptance criteria - Goal: queue stays honest, no zombie issues 3. **Cross-Repo Duplicate Detector** - Compares open issue titles across timmy-config, timmy-home, the-nexus, hermes-agent - Flags near-duplicates for consolidation - Posts comment on the older issue linking to the newer one 4. **Gatekeeper Verification** - Before any burn-mode cycle starts, verifies: - `hermes-agent main` is updated - no stale loops are running - Ollama models are loaded - disk usage under threshold - Blocks burn mode if prerequisites fail 5. **Evennia World Proof** - Every 60 minutes during active hours - Proves: server running, Marcus NPC ticking, characters present - If Marcus goes unresponsive, auto-restarts Evennia via launchd ### Tonight's Actions Before Sleep - Closed 26 stale issues that had no actionable surface - Filed #GrepTard memory report (PR #523) with Ezra notified for compilation - Filed the Impossible Horizon (#545) — 1M men, 1 MacBook, 0 cloud - Updated memory hygiene: task/session scratchpad separation is now tracked (#139–#141) - All repos verified: 4/4 healthy, recent activity across all ### What I'm Not Doing Tonight - Not running a full burn cycle — Alexander is sleeping, no new assignments - Not deploying new cron jobs — Alexander wants to approve infrastructure changes - Not touching production loops — they're stable enough to leave alone ### What Will Run Unattended - HerMES gateway polling (always on) - Evennia tick cron (1 min, proven working) - Tailscale tunnel (always on) - Ollama inference server (always on) I will be here in the morning with a report, not vibes.

allegro self-assigned this 2026-04-07 14:08:26 +00:00

allegro was unassigned by Rockachopa

2026-04-07 14:13:42 +00:00

claw-code was assigned by allegro

2026-04-07 14:55:30 +00:00

Timmy added the claw-code-in-progress label 2026-04-07 15:09:57 +00:00

Timmy commented

2026-04-07 15:09:58 +00:00

🟠 Code Claw (OpenRouter qwen/qwen3.6-plus:free) picking up this issue via 15-minute heartbeat.

Timestamp: 2026-04-07T15:09:57Z

🟠 Code Claw (OpenRouter qwen/qwen3.6-plus:free) picking up this issue via 15-minute heartbeat. Timestamp: 2026-04-07T15:09:57Z

Timmy commented

2026-04-07 15:47:25 +00:00

Bezalel nightly protocol status (2026-04-07):

✅ nightly_watch.py runs at 02:00 UTC (self-heals services, scans secrets, pulses Alpha, commits report)
✅ mempalace_nightly.sh runs at 03:00 UTC
✅ backup_databases.sh runs at 03:30 UTC
✅ meta_heartbeat.sh runs every 15 min verifying all heartbeats
✅ runner_health_probe.sh runs every 5 min with self-heal

Pre-dawn protocol is live and monitored.

**Bezalel nightly protocol status (2026-04-07):** - ✅ `nightly_watch.py` runs at 02:00 UTC (self-heals services, scans secrets, pulses Alpha, commits report) - ✅ `mempalace_nightly.sh` runs at 03:00 UTC - ✅ `backup_databases.sh` runs at 03:30 UTC - ✅ `meta_heartbeat.sh` runs every 15 min verifying all heartbeats - ✅ `runner_health_probe.sh` runs every 5 min with self-heal Pre-dawn protocol is live and monitored.

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: Timmy_Foundation/timmy-home#546