skills.rest
burningportra
burningportraCOMMUNITY
Explore all skills by this author →
GitHub

flywheel-swarm-stop

Stops all running swarm agents and releases their file reservations.

3|1|Updated Apr 15, 2026
Search Tags:#claude code#multi-agent#cleanup#swarm#shutdown#agent-mail#file reservations
One-click install
npx skills add https://github.com/burningportra/agent-flywheel-plugin --skill flywheel-swarm-stop-burningportra
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: flywheel-swarm-stop
Source: https://github.com/burningportra/agent-flywheel-plugin/tree/main/skills/flywheel-swarm-stop
Command: npx skills add https://github.com/burningportra/agent-flywheel-plugin --skill flywheel-swarm-stop-burningportra

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Shutting down a multi-agent coding swarm cleanly is error-prone: agents may hold file reservations, leave beads stuck in an in_progress state, and ignore broadcast stop signals. This Skill performs an orderly teardown so no work is lost and no locks are left behind. ## Core Features & Use Cases - Graceful agent shutdown: Identifies active agents via the agent-mail MCP inbox and sends each one an individual urgent stop message followed by a structured shutdown request. - Resource cleanup: Releases all file reservations for the current project and resets in-progress beads back to open so work can be resumed later. - Task reconciliation: Marks active swarm tasks as deleted and reports how many agents were signaled and beads reset. - Use Case: After a multi-agent implementation wave finishes or needs to be aborted, run this Skill to stop every agent, free reserved files, and restore the bead queue to a clean state. ## Quick Start Stop all running swarm agents, release their file reservations, and reset any in-progress beads to open.

Frequently Asked Questions about flywheel-swarm-stop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stop all running swarm agents in Claude Code?▼

Fetch active agents via the agent-mail fetch_inbox tool, then send each agent an individual urgent stop message followed by a structured shutdown_request. Use TaskStop with saved task IDs for agents that do not respond.

How do I release file reservations after stopping a swarm?▼

Call release_file_reservations through the agent-mail MCP tool with project_key set to the current working directory. This frees all file locks held by swarm agents so other work can proceed without conflicts.

Why can't I broadcast a stop message to all agents at once?▼

Structured messages in agent-mail cannot be broadcast to "*", so each active agent must be messaged individually by name. Send a plain-text urgent nudge first, then the structured shutdown_request payload.

What happens to in-progress beads when the swarm stops?▼

Beads still marked in_progress are reset to open using br update <id> --status open. This returns unfinished work to the queue so it can be picked up in a future session.

What if a swarm agent ignores the shutdown request?▼

Use TaskStop with the task ID saved when the agent was spawned to force termination. Then use TaskList to find remaining active swarm tasks and mark each as deleted via TaskUpdate.

Related Skills

Explore More Software Engineering Skills→
DicklesworthstoneDicklesworthstone
70

agent-swarm-workflow

Coordinate parallel software tasks across multiple agents using Agent Mail and BV prompts.

Community
Advanced
laicluselaicluse
4

wake

Resume a stalled autonomous loop from its loop file and restart its cron heartbeat.

Official
Intermediate
justin-delanojustin-delano
1

beads-audit

Run Beads toolkit health checks to detect orphaned issues and circular dependencies.

Community
Intermediate
powdreampowdream

stop-emulator

Stops a specified Android emulator (AVD) by name or serial using adb.

Community
Intermediate