🔊 OpenClaw Lean VPS — Audio Summary
Research findings on running OpenClaw on a 1 vCPU / 2 GB RAM VPS
Generated by gemAudio · Zephyr voice
Key takeaways:
- One channel only — Telegram recommended (20–40 MB RAM)
- Sandbox OFF — doubles gateway memory footprint
- cron.maxConcurrentRuns = 1
- History limits ≤ 20 messages
- Node heap: 1536 MB gateway / 768 MB subagents
- Avoid deep-research, browser automation on 2GB