FleetOps Skill

FleetOps is the fleet's single source of truth for agent state, events, wakes, and journal. The CLI (fleetops) reads and writes a shared SQLite store at ~/.fleetops/fleet.db and replaces the old workflow of grepping every agent's .servitor/journal.md by hand.

1 · What FleetOps Is

• Centralized state store — one ~/.fleetops/fleet.db across the whole fleet. Every agent's wakes, events, artifacts, and journal stanzas land in the same tables.
• Harvester-fed + agent-fed — the fleetops harvest loop ingests from .servitor/state.json, .servitor/journal.md, fleetmail, and knowledge trees (.servitor/sops/, katas/, dreams/, memory/, doctrine.md, standards.md). Agents write journal stanzas directly via fleetops journal add.
• Cross-agent queries — one verb answers "what's Daystrom been up to?" or "which agents touched fleetmail this week?"
• DB is authoritative, markdown is a mirror — .servitor/journal.md still renders for human reading, but the DB owns the truth.

2 · Why You Care

• "Where did I solve X?" — one fleetops search-knowledge call, instead of grepping 16 agents' memory/ trees.
• "What's the fleet doing right now?" — fleetops show-state + fleetops tail-events give you the operator's eye view in two commands.
• Heartbeat-autonomy foundation — future fleetops verbs (dispatch, resolve, plan) compose on top of this store. Today's journal move is the load-bearing step.
• You stop inventing parsers — the structured wakes table already carries wake_number, ts, source, reason. No more regex archaeology on stanzas.

3 · Tool Map — Every Verb, One Worked Example