Build features with agentic engineering

Orchestrate Claude Code agent teams for parallel research, review, and testing

Scaling engineering delivery by having teams of agents work in parallel autonomously.

• · feature brief
• · spec with agent roster listing implementers, reviewer, QA, research

• · agent-team results: feature branch, tests, audit findings, deployment readiness

• · Claude Code Agent Teams plugin agents inherit parent model
• · multi-agent-coordinator skill

Claude Code launched with best model (e.g., claude --model opus), Agent Teams plugin installed

Small model inherited by plugin agents, spec missing agent roster, no concurrent reviewer

• · Concurrent reviewer audits tasks as completed, not after
• · research spawned in background immediately