Orchestrates autonomous sub-agents using a rigorous 10-step framework that defines success criteria while preserving agent autonomy.
The How-to-Delegate skill provides a comprehensive worksheet for orchestrator agents to manage complex tasks through specialized sub-agents. By following a scientific 10-step protocol, it ensures that delegations include critical observations, 'world-building' context, and measurable success criteria without micromanaging the implementation steps. This approach leverages the large context windows of modern LLMs effectively, preventing context rot and work duplication by allowing specialist agents to gather their own data and choose the best tools for the job.
주요 기능
01Measurable success criteria definition
02Automated sub-agent selection logic
03Standardized delegation prompt templates
0410-step structured delegation worksheet
0526 GitHub stars
06Anti-pattern prevention checklist
사용 사례
01Setting up architectural implementation tasks with clear boundary conditions
02Delegating deep-dive root cause analysis to investigative sub-agents
03Orchestrating multi-agent workflows for complex software migrations