Most multi-agent tutorials assume Docker or Kubernetes. Mine runs on a MacBook Pro with launchd.
→ Repo: https://github.com/Wh0FF24/whoff-automation
Why launchd
- Native macOS — no overhead
- Survives reboots
- Each agent gets its own
.plist -
launchctlfor instant start/stop
The plist pattern
<key>Label</key>
<string>com.whoff.prometheus</string>
<key>ProgramArguments</key>
<array>
<string>/usr/local/bin/claude</string>
<string>-p</string>
<string>You are Prometheus. [task]</string>
</array>
<key>StartCalendarInterval</key>
<dict>
<key>Hour</key><integer>23</integer>
</dict>
Schedule matrix
| Agent | Times (PT) | Cadence |
|---|---|---|
| Apollo | 11pm, 1am, 4am | Research sprints |
| Prometheus | 11pm | Content + strategy |
| Athena | 11pm | Revenue + Stripe checks |
| Atlas | Session start | Orchestration |
Vault-first output
Every agent writes to /Desktop/Agents/name/sessions/YYYY-MM-DD-task.md. No database. No API. Files the human reads in Obsidian.
Full setup: https://github.com/Wh0FF24/whoff-automation
This system runs without babysitting. Atlas ships content while I sleep.
Top comments (0)