launch-swarm

Launch a parallel Claude Code agent swarm for labeled beads via ntm.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/quantfiction/dotfiles --skill launch-swarm
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: launch-swarm
Source: https://github.com/quantfiction/dotfiles/tree/main/claude/plugins/global-skills/skills/launch-swarm
Command: npx skills add https://github.com/quantfiction/dotfiles --skill launch-swarm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Launch a parallel Claude Code agent swarm to work on beads that share a specific label, coordinating models, templates, and execution to accelerate bead processing.

Core Features & Use Cases

  • Spawns multiple agent panes using ntm to accelerate bead processing.
  • Sends bead_swarm templates to all agents and monitors progress across the swarm.
  • Uses full model names to avoid alias issues and ensures deterministic agent behavior.

Quick Start

Provide the bead label to initiate a swarm that processes beads labeled accordingly.

Frequently Asked Questions about launch-swarm

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

FAQPage Schema
How do I launch a parallel Claude Code agent swarm to work on beads?

To launch a parallel Claude Code swarm, the skill spawns multiple agent panes using ntm to manage tmux sessions, sending bead_swarm templates to process beads that share a specific label.

Why should I use full model identifiers when spawning Claude Code agents in tmux?

Using full model identifiers when spawning Claude Code agents avoids version alias problems, ensuring deterministic agent behavior across the swarm without unexpected model mismatches.

What prerequisites do I need to orchestrate a Claude Code swarm for beads?

Orchestrating a Claude Code swarm requires ntm to manage tmux sessions and prior completion of BEADS.md creation and bead QA steps before spawning agents.

How can I monitor progress across multiple Claude Code panes in a swarm?

After spawning the swarm, the skill verifies execution across panes and exposes monitoring and control commands to track agent progress and manage the tmux sessions.

What is the best way to automate parallel bead processing with Claude Code?

Automating parallel bead processing is best done by orchestrating a Claude Code swarm, which coordinates models, sends templates to all agents, and monitors execution across panes.

Are there limitations when using ntm to manage Claude Code swarm sessions?

A key limitation is that the swarm relies on ntm to manage tmux sessions, meaning execution and verification across panes depend on tmux session stability and proper ntm configuration.