An operating system for one person running a lot of things.
lifeOS is the dashboard at claw.philgetzen.com. Five tabs — Home, Briefings, Automations, Agents, Wikis — pulling from three layers: an autonomous agent system, a planning database, and a nine-wiki knowledge vault.
Not a product for others. This is the infrastructure that actually runs a consulting practice and a portfolio of products. It's worth showing because it's the test bed for everything else, and because it works.
What's underneath the dashboard
Agent Layer
ClaudeClaw runs the recurring ops work, autonomously.
Claude Code plugin with a persistent background daemon and cron-style scheduling. A Haiku watchdog polls and decides what escalates to a flagship model. Notifications route by urgency, not by source. Three agent groups cover executive assistant, BookOwl, and consulting tasks.
- Orchestrator auto-discovers agent groups from config — no hardcoding
- Per-agent trust tiers: autonomous, notify, or approval-required
- Watchdog on Haiku keeps polling cheap; flagship models run only when invited
- Scheduled outputs write directly into the wiki vault raw directories for synthesis
Planning Layer
PG Dash is the system of record every other tool reads from.
Vite admin dashboard on top of a Firestore schema covering epics, items, and milestones across 18 workspaces. The same data is exposed as an MCP server, so Claude Code, ClaudeClaw, and Product Partner read the live roadmap directly without copying anything.
- Workspace, project, epic, item hierarchy built for multi-venture planning
- MCP server with 20+ read and selective write tools for agent clients
- Schema contract locked across PG Dash and Product Partner's firebase adapter
- Activity feed and cleanup scripts keep the planning layer from drifting
Briefings
Every agent run lands somewhere readable.
Morning briefings, end-of-day digests, weekly planning prose, and automation output all route to the Briefings tab. One place to read what happened without digging through logs or notifications.
- Briefings separated from automation output — narrative prose vs. raw results
- IMPORTANT / FYI priority tiers keep high-signal items visible
- Readable summaries generated by flagship models, not raw agent logs
- Archive of every run output — searchable, timestamped, expandable
Automations
Cloud routines and local daemon schedules in one place.
Every scheduled task — cloud-hosted or running on the local daemon — is visible, manually triggerable, and linked to its output in Briefings. Executive assistant, BookOwl, and consulting automations are grouped by agent.
- Cloud vs. local schedule type shown per automation row
- Manual run triggers directly from the dashboard without touching config
- Last run status and timestamp visible at a glance
- Grouped by agent — Executive Assistant, BookOwl, Consulting — not by schedule
Memory Layer
Nine interconnected wikis synthesize everything into something readable.
Raw sources — Claude conversations, PDFs, LinkedIn exports, source repos — land in raw/ and get synthesized into structured _index.md pages. ClaudeClaw agents write scheduled outputs straight into the right raw/ directory. Downstream tools read the synthesized layer, not the mess underneath.
- Nine wikis: writing, BookOwl, self, tech, reading, career, networking, work-history, consulting
- Raw inputs and synthesized output live in separate folders, never mixed
- CLAUDE.md schema in the vault so any model knows where things belong without being told
- Editable from a phone via iCloud sync; pages cross-link across wikis
System
One map of every moving part, live.
The System tab shows the full infrastructure state: daemon uptime, node health, task failures, wiki sync status, and Claw Ops state. Thirteen nodes tracked, one surface to see when something needs attention.
- Node health at a glance — green dots for healthy, gray for stale, orange for flagged
- NEEDS A LOOK panel surfaces failures before they become problems
- Daemon uptime and last-run metadata without digging into logs
- Wiki sync state and raw file counts pulled live from the vault
How a solo operator runs a lot of things
Three layers, one dashboard, and enough automation that the recurring work mostly takes care of itself.