What problem does it solve? Deciding how many agents to spawn, which roles to assign, and which subagent_type each teammate needs is error-prone: read-only agents get assigned implementation work, teams grow too large, and review dimensions overlap. This Skill provides sizing heuristics, preset team configurations, and agent type selection rules for Claude Code's Agent Teams feature. ## Core Features & Use Cases - Team Sizing Heuristics: Match team size (1-5 agents) to task complexity, from single-dimension reviews to full-stack features. - Preset Team Compositions: Ready-made configurations for review, debug, feature, fullstack, research, security, and migration teams with defined roles and dimensions. - Agent Type Selection: Decision matrix mapping roles to subagent_type values (general-purpose, Explore, Plan, team-reviewer, team-debugger, team-implementer, team-lead) based on tool access needs. - Display Mode Configuration: Guidance for tmux, iTerm2, and in-process teammate display modes for local and CI environments. - Use Case: When planning a security audit, use the Security Team preset to spawn four reviewers covering OWASP vulnerabilities, auth/access control, dependencies, and secrets/configuration. ## Quick Start Ask the agent to design a team composition for your task, for example: "Set up a debug team with three investigators to find the root cause of this failing test."
