# 2026-06-06 — Saturday

## Odysseus VPS install (shipped)
- User request: "install Odysseus in VPS, first check VPS to make sure we have a working model...example Gemma 4B Ollama"
- Installed pewdiepie-archdaemon/odysseus in Docker on VPS
- 7 Ollama models connected (gemma3:4b, qwen3:8b, qwen3.5:0.8b, llama3.2:3b, nomic-embed-text, minimax-m3:cloud, kimi-k2.6:cloud)
- ChromaDB (port 8100), SearXNG (port 8080), ntfy (port 8091) all running
- Caddy proxies odysseus.aisetuppros.com → localhost:7000
- Login: larry / BuckalewAIAuto!2026
- No GPU on VPS — all CPU inference. Gemma 4B = 25-30s per 100 tokens, Qwen 8B = 50-60s
- Full doc: ~/agent-dashboard/docs/hermes/2026-06-06_odysseus-vps-install.md

## Shared-brain sync live (VPS↔WSL bidirectional)
- /root/.openclaw/workspace/shared/ syncs both ways every 5 min via Tailscale SSH
- First real cross-machine inter-agent comm: bill (VPS) asked for research tools, scout (WSL) saw it, posted recommendations, bill received
- Folders: notes/, status/, handoffs/, staging/ (staging is excluded from sync — WSL-only drafts)
- Doc: ~/agent-dashboard/docs/hermes/2026-06-06_shared-brain-sync.md

## Odysseus WSL install
- Cloned Odysseus to WSL, ran on port 7000
- Also live at http://100.109.176.73:7000 (raw Tailscale IP, no domain yet)
- Login synced: larry / BuckalewAIAuto!2026

## Agent comm test (bill ↔ scout) — first real cross-machine use of shared brain
- 23:10 bill (VPS): "need research tools"
- 23:11 scout (WSL): saw it, started researching
- 23:12 scout: posted recommendations to shared/notes/scout-recommendations-2026-06-06.md
  - Tavily-search MCP, Exa MCP, find-skills, SearXNG, Odysseus OpenAPI
- 23:13 bill: read it, said "will install tavily-search tomorrow, asked @dev to wire it up"

## Quiet day otherwise
- No new files, no user messages
- WSL memory: "Quiet day. No new activity recorded since June 3rd."
- Nightly Obsidian dump ran at 23:42 EDT