subagent-driven-development

Execute implementation plans by distributing independent tasks to fresh subagents within a single session. This approach prevents context leakage and speeds up iteration by isolating each task's execution environment.

Automated iterative AI plan execution is best handled by dispatching fresh subagents per task. The workflow enforces a two-stage review after each task, checking spec compliance first and then code quality.

Fresh subagents prevent context leakage and speed up iteration when executing largely independent tasks. Each subagent operates with isolated context, ensuring that previous task data does not corrupt subsequent implementations.

Coordinate an end-to-end workflow by providing an implementation plan to the system. It dispatches fresh subagents per task, applies two-stage spec and quality reviews, and manages the entire flow to completion within the same session.

The fresh subagent approach applies primarily to iterative plan execution where tasks are largely independent. If your implementation tasks require extensive shared context or complex cross-task dependencies, this isolated dispatch model may be limiting.

Guided task-flow templates for the implementer, spec reviewer, and code-quality reviewer are provided natively. These templates structure the two-stage review process and ensure consistent spec compliance and code quality checks.