Files
timmy-home/configs/phase-1-snapshot.json

37 lines
1.7 KiB
JSON

{
"fleet_operational": true,
"resources": {
"uptime_percent": 78.0,
"days_at_or_above_95_percent": 0,
"capacity_utilization_percent": 35.0
},
"current_buildings": [
"VPS hosts: Ezra (143.198.27.163), Allegro, Bezalel (167.99.126.228)",
"Agents: Timmy harness (local Mac M4), Code Claw heartbeat, Gemini AI Studio worker",
"Gitea forge at forge.alexanderwhitestone.com (16 repos, 500+ issues)",
"Ollama local inference (6 models, ~37GB)",
"Hermes agent (cron system, 90+ jobs, 6 workers)",
"Tmux fleet (BURN session, 50+ panes)",
"Evennia MUD worlds (The Tower, federation)",
"RunPod GPU pod (L40S 48GB, intermittent)"
],
"manual_clicks": [
"Restart agents and services by SSH when a node goes dark",
"Check VPS health (disk, memory, process) via manual SSH",
"Verify Gitea, Ollama, and Evennia services after deployments",
"Merge PRs manually \u2014 auto-merge covers ~80%, rest need human review",
"Recover dead tmux panes \u2014 no auto-respawn wired yet",
"Handle provider failover \u2014 no automated switching on OOM/timeout",
"Triage the 500+ issue backlog \u2014 burn loops help but need supervision",
"Run nightly retro and push results to Gitea"
],
"notes": [
"Fleet is operational but fragile \u2014 most recovery is still manual",
"Overnight burns work ~70% of the time; 30% need morning rescue",
"The deadman switch exists but is not in cron (fleet-ops#168)",
"Heartbeat files exist but no automated monitoring reads them",
"Provider failover is manual \u2014 Nous goes down = agents stop",
"Phase 2 trigger requires 30 days at 95% uptime \u2014 we are at 0 days"
],
"last_updated": "2026-04-14T22:00:00Z"
}