Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From memfleet-skills
Use when orchestrating a multi-step fleet edit involving multiple agents, multi-symbol refactors, or when you are the leader agent coordinating followers. Triggered by: 'coordinate agents on this refactor', 'split this work across the fleet', 'watch the fleet while I drive a large change', leader-follower patterns.
npx claudepluginhub syncable-dev/memfleet-public --plugin memfleet-skillsHow this skill is triggered — by the user, by Claude, or both
Slash command
/memfleet-skills:memfleet-fleet-coordinationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Large refactors that touch many symbols benefit from a leader/follower pattern. The leader publishes a coordinated set of intents; followers subscribe and pick work off the propagation set.
Orchestrates multiple parallel work streams in isolated worktrees across coordinated waves. Use when work decomposes into 3+ independent tasks.
Manages a sequential landing queue for parallel git branches: test gate, pre-land scrub, auto-rebase, and one-shot revert. Use after spawning parallel sessions with native agent teams or background agents.
Guides multi-agent coordination in OpenCode swarm workflows with decomposition, worker spawning, file reservations, progress tracking, and review loops. Use for parallelizable tasks across 3+ files.
Share bugs, ideas, or general feedback.
Large refactors that touch many symbols benefit from a leader/follower pattern. The leader publishes a coordinated set of intents; followers subscribe and pick work off the propagation set.
1. fleet_status(window_ms=1_800_000) — confirm the fleet is idle enough
2. For each planned symbol, publish_intent with a common correlation tag
3. Announce the correlation tag on the shared MCP channel (or via a single "kickoff" episode)
4. Subscribe with the correlation tag filter
5. Wait for followers' record_episode events → aggregate class, retry losers, escalate C
1. Subscribe to { intent_kinds: [Refactor, FeatureAdd], modules: [<area>] }
2. On notification, if correlation_tag matches → claim with a publish_intent of your own on the leaf symbol
3. If claim succeeds → memfleet-safe-edit on that symbol
4. Report back via record_episode with the correlation tag
Since Phase-1 does not have a native tag field, encode the tag in intent_kind's free-form surface (e.g. FeatureAdd { surface: NewField("coord=<ulid>") }) until Phase-2 adds it.
resolve_conflict (Phase-2) rather than letting partial state land.record_episode uses the same intent_id it claimed with; replays are no-ops.Subscriptions are stub. In practice today, leader + followers poll query_episodes with a since cursor and filter by the correlation tag embedded in the intent surface. Upgrade to real push subscriptions in Phase-2.