add-subagent-driven-development

Automated code reviews between tasks are enforced by dispatching independent subagents for each discrete task, catching issues early before the next iteration begins. This prevents context pollution and maintains quality across complex feature implementations.

Subagent-driven development orchestrates iterative software execution by breaking down plans into atomic tasks and dispatching fresh subagents for each one. It requires a robust dispatch mechanism and structured decision logging to maintain context and track progress across tasks.

Task decomposition breaks down a complex implementation plan into atomic, executable tasks suited for independent subagent execution. Each task is then orchestrated sequentially with integrated quality gates to ensure rapid iteration and high code quality.

Subagent orchestration applies to feature implementation workflows requiring coordinated multi-agent execution, such as creating database migrations, backend API endpoints, and frontend components. It manages execution of each part while enforcing code reviews at every step.

A structured decision logging system is required to maintain context and track progress across tasks during subagent-driven development. It ensures that independent subagents remain coordinated throughout the iterative execution and quality gate cycles.

Subagent-driven development should be avoided when a feature cannot be decomposed into discrete tasks or when a robust subagent dispatch mechanism is unavailable. It is designed for complex, multi-step implementations rather than simple, single-action code changes.