build-with-agent-team

Multi-agent coordination works by analyzing a structured plan to determine team size, assign roles, and spawn Claude Code agents in parallel. This skill enforces exact interfaces, validates contracts, and provides step-by-step guidance for orchestration and end-to-end validation.

Contract-first coordination establishes role ownership and cross-cutting concerns up front before agents start building. This skill validates these contracts early, enforcing exact interfaces between agents to prevent integration failures during the multi-agent build process.

You provide a detailed plan document and run this skill to spawn a Claude Code Agent Team. It forms parallel teams from the plan, handles delegate-mode spawn, and provides real-time coordination guidance for both greenfield and brownfield projects.

Yes, tmux is supported for managing parallel agent coordination. The skill provides coordination guidance for delegate-mode spawn and real-time orchestration, allowing you to monitor and manage multiple Claude Code agents building simultaneously across tmux sessions.

Yes, a detailed plan document is required. The skill analyzes this plan to determine team size, ownership boundaries, contracts, and cross-cutting concerns, then uses that structure to spawn and coordinate agents in parallel for the build.

You should avoid multi-agent builds when your project lacks a clear structured plan or defined interfaces between components. Without upfront contracts and role ownership to validate, parallel agents risk producing incompatible code that fails end-to-end validation.