Agent Swarm Studio

## Overview
Agent Swarm Studio helps you turn a messy, multi-track task into a coordinated agent swarm. Use it to define clear agent roles, split independent work into parallel tracks, and keep the swarm converging on a usable result instead of drifting into disconnected outputs.

## Quick Start
1. Start with `dnyoussef-agent-creator` (`agent-creator`) to define each specialized agent's role, system prompt, responsibility boundary, and domain knowledge.
2. Use `dmjgilbert-parallel-agents` (`parallel-agents`) when the work has 3 or more independent failures, investigations, or task tracks that can run concurrently.
3. Bring the outputs together with `dnyoussef-swarm-orchestration` (`swarm-orchestration`) to coordinate the full lifecycle, route work through the right topology, and check that the swarm converges before handoff.

## Key Commands
- `dnyoussef-agent-creator` / `agent-creator` - create production-ready specialized agents with optimized system prompts and clear responsibility boundaries.
- `dmjgilbert-parallel-agents` / `parallel-agents` - dispatch multiple agents across independent problems so work can happen concurrently.
- `dnyoussef-swarm-orchestration` / `swarm-orchestration` - orchestrate multi-agent swarms for complex workflows, parallel execution, dynamic topology, and intelligent coordination.

## Tips
- Use Agent Swarm Studio for task planning, incident triage, research synthesis, codebase migration planning, or any workflow where one agent would become a bottleneck.
- Keep the swarm small at first: create only the roles you can clearly name, parallelize only independent tracks, and add a convergence checkpoint before final delivery.
- Before launching parallel work, write the expected merge format so each agent returns comparable output.