agent-orchestration

What problem does it solve? Deciding which AI model should handle a task, how much reasoning effort to apply, and how to split work across multiple agents is error-prone and inconsistent. This Skill provides a structured decision framework for delegation so work is routed to the right model with the right role and effort every time. ## Core Features & Use Cases - Model and Role Selection: Ranks models by intelligence, taste, and cost, and maps them to Planner, Builder, Reviewer/Judge, and cheap-worker roles with explicit fallback orders. - Delegation Patterns: Supports specialist, fan-out, staged plan/build/review/fix, and second-opinion patterns with rules for parallelization and retry limits. - Delegation Packets: Defines a structured packet format covering objectives, success criteria, constraints, style rules, and verification steps so delegates receive complete context. - Use Case: When implementing a feature, use this Skill to route planning to a high-intelligence model, building to a cost-effective Builder, and review to an independent cross-family Reviewer, then coordinate fixes through staged phases. ## Quick Start Ask the agent to decide whether this task should be delegated and to which model, role, and effort level using the agent-orchestration skill.