multi-agent-coordination

What problem does it solve?

Coordinating multiple agents—whether in parallel or sequentially—can be challenging due to dependencies, task routing, and context management. This Skill provides patterns to orchestrate specialized agents, delegate to sub-agents, and manage context windows to maximize speed and reliability.

Core Features & Use Cases

• Execution Patterns: Choose parallel vs sequential execution based on dependencies.
• Agent Selection & Task Typing: Map tasks to appropriate agents to optimize outcomes.
• Sub-Agent Delegation: Use file-based instructions to isolate contexts and improve debuggability.
• Context Window Management: Strategies to preserve relevant information across calls.
• Scenario Examples: Parallel validation from multiple viewpoints; multi-phase design with dependent steps.

Quick Start

To begin coordinating agents:

1. Decide between parallel or sequential workflow and assign tasks to the appropriate agents (e.g., designer, backend-developer, tester).
2. If delegating to sub-agents, place instructions in a file (e.g., ai-docs/delegation-instructions.md) and reference it from the task prompt.
3. Start the workflow and monitor outputs; consolidate results after completion.