Running profClaw on Raspberry Pi Zero: what actually works

[thegdsks]

Tested pico mode on a Pi Zero 2W (512MB RAM, ARM64). Here is what we found.

npm install -g profclaw gets OOM-killed every time. The Pi just doesn't have enough memory for npm to compile native deps.

What works:

1. Docker pico image. 100MB RAM usage. Full agent engine, 72 tools, CLI, API.

docker run -d -p 3000:3000 -e PROFCLAW_MODE=pico ghcr.io/profclaw/profclaw:latest

2. Add 1GB swap before npm install. Slow (SD card) but it finishes.

sudo dphys-swapfile swapoff
sudo sed -i 's/CONF_SWAPSIZE=.*/CONF_SWAPSIZE=1024/' /etc/dphys-swapfile
sudo dphys-swapfile setup && sudo dphys-swapfile swapon
npm install -g profclaw

3. For local AI, point at Ollama running on a beefier machine:

-e OLLAMA_BASE_URL=http://your-pc:11434

Pico mode has no UI dashboard and no Redis. Everything else works: all 72 tools, CLI chat, REST API, cron jobs, one messaging channel.

Anyone else running on SBCs? Curious what hardware people are using.