super-swarm-spark
Run 12 Parallel AI Agents on Your Dev Plan
Executing large development plans sequentially is slow. This skill orchestrates up to 12 concurrent AI subagents that work in parallel on different tasks, dramatically speeding up project completion while maintaining coordination.
Descargar el ZIP de la skill
Subir en Claude
Ve a Configuración → Capacidades → Skills → Subir skill
Activa y empieza a usar
Pruébalo
Usando "super-swarm-spark". /super-swarm-spark auth-plan.md
Resultado esperado:
Starting parallel execution with 12 Sparky agents...
Launched: T1 (User Model), T2 (Auth Controller), T3 (JWT Handler)...
Task T1 completed: Created User model with validation
Task T2 completed: Implemented auth endpoints
Integration pass: Resolved 2 path conflicts
Final status: 8/8 tasks completed, all tests passing
Usando "super-swarm-spark". /super-swarm-spark ./plans/api-plan.md T1 T2 T4
Resultado esperado:
Running subset: T1, T2, T4
Launched 3 concurrent agents...
All tasks completed successfully.
Updated plan file with completion logs.
Auditoría de seguridad
Riesgo bajoAll 22 static findings are false positives. The scanner misinterpreted markdown backticks as shell execution commands and generated spurious crypto algorithm warnings. This skill is a legitimate agent orchestrator that manages parallel subagent execution for development tasks. No malicious behavior detected.
Problemas de riesgo alto (1)
Problemas de riesgo medio (1)
Problemas de riesgo bajo (1)
Factores de riesgo
⚙️ Comandos externos (10)
Puntuación de calidad
Lo que puedes crear
Accelerate Large Feature Implementations
When implementing a large feature that can be broken into independent tasks, use this skill to run multiple subagents in parallel, reducing total execution time significantly.
Parallelize Test Writing
Run multiple subagents simultaneously to write tests for different modules or components of a codebase in parallel.
Batch Refactoring Tasks
Execute multiple refactoring tasks across different parts of the codebase concurrently while maintaining naming consistency through context packs.
Prueba estos prompts
Execute the plan file ./my-project-plan.md using super-swarm-spark. Run all tasks in the plan.
Execute tasks T1, T3, and T5 from ./backend-plan.md using super-swarm-spark.
Continue executing ./feature-plan.md. The following tasks are already complete: T1, T2. Run the remaining pending tasks.
After all subagents complete their tasks in ./full-plan.md, perform the final integration pass and run tests to ensure everything works together.
Mejores prácticas
- Ensure plan files have clear task IDs (T1, T2, etc.) and complete acceptance criteria for each task
- Provide canonical file paths in the plan to help the orchestrator prevent naming conflicts
- Keep tasks relatively independent to maximize parallel execution benefits
Evitar
- Do not use this skill for plans with heavy dependencies between tasks (subagents will block waiting on each other)
- Avoid running this skill without a properly formatted plan file
- Do not expect the orchestrator to handle tasks requiring the same file simultaneously