What problem does it solve? Breaking large features into parallelizable work for agent teams is error-prone: unclear ownership causes file conflicts, missing dependencies cause deadlocks, and uneven workloads leave teammates idle. This Skill provides structured strategies for task decomposition, dependency graph design, and workload monitoring. ## Core Features & Use Cases - Task Decomposition Strategies: Split work by layer, component, concern, or file ownership to maximize parallelism and avoid merge conflicts. - Dependency Graph Design: Apply patterns like independent, sequential, diamond, fork-join, and pipeline graphs using blockedBy/blocks relationships, while avoiding circular dependencies and bottlenecks. - Task Description Templates: Write tasks with objectives, owned files, interface contracts, acceptance criteria, and explicit scope boundaries. - Workload Monitoring: Detect imbalance signals (idle teammates, blocked tasks, overload) and rebalance using TaskList, TaskUpdate, and SendMessage. - Use Case: When building a full-stack authentication feature with three agents, decompose it into login, registration, and shared-infrastructure streams, wire dependencies so implementation waits on shared types, and monitor progress to reassign work when one stream finishes early. ## Quick Start Ask the agent to break down your feature into parallel tasks with clear file ownership and dependency relationships for a team of agents.
