- 8-agent OpenClaw cluster (Manager, Tech-Lead, Secretary, Auditor, Optimizer, Study-Builder, NX-Expert, Webster) - Orchestration engine: orchestrate.py (sync delegation + handoffs) - Workflow engine: YAML-defined multi-step pipelines - Agent workspaces: SOUL.md, AGENTS.md, MEMORY.md per agent - Shared skills: delegate, orchestrate, atomizer-protocols - Capability registry (AGENTS_REGISTRY.json) - Cluster management: cluster.sh, systemd template - All secrets replaced with env var references
Windows Setup — Atomizer Job Queue
Quick Setup
- Copy this folder to
C:\Atomizer\on Windows - Create the job queue directories:
mkdir C:\Atomizer\job-queue\pending mkdir C:\Atomizer\job-queue\running mkdir C:\Atomizer\job-queue\completed mkdir C:\Atomizer\job-queue\failed - Set up Syncthing to sync
C:\Atomizer\job-queue\↔/home/papa/atomizer/job-queue/ - Edit
atomizer_job_watcher.py— updateCONDA_PYTHONpath if needed
Running the Watcher
Manual (recommended for now)
conda activate atomizer
python C:\Atomizer\atomizer_job_watcher.py
Process single pending job
python C:\Atomizer\atomizer_job_watcher.py --once
As a Windows Service (optional)
# Install NSSM: https://nssm.cc/
nssm install AtomizerJobWatcher "C:\Users\antoi\anaconda3\envs\atomizer\python.exe" "C:\Atomizer\atomizer_job_watcher.py"
nssm set AtomizerJobWatcher AppDirectory "C:\Atomizer"
nssm start AtomizerJobWatcher
How It Works
- Agents on Linux write job directories to
/job-queue/outbox/ - Syncthing syncs to
C:\Atomizer\job-queue\pending\ - Watcher picks up new jobs, runs them, moves to
completed/orfailed/ - Results sync back to Linux via Syncthing
- Agents detect completed jobs and process results
Note
For Phase 0, Antoine runs python run_optimization.py manually instead of using the watcher.
The watcher is for Phase 1+ when the workflow is more automated.