subagent-driven-development

Automate plan execution by dispatching a fresh subagent per task, which isolates context and enables parallel work. This multi-agent approach prevents context leakage and drives fast, quality-focused development across independent tasks.

A fixed multi-agent workflow coordinates a central controller that dispatches per-task implementer subagents, followed by spec-reviewer and code-quality-reviewer subagents. This enforces spec compliance and quality review for autonomous development.

Set up two-stage reviews by applying template prompts for spec-reviewer and code-quality-reviewer subagents. After an implementer subagent completes a task, the workflow automatically triggers these dedicated reviews to ensure high-quality results.

Use a fresh subagent per task when your autonomous development workflow involves relatively independent tasks. This approach is ideal when you need fast iteration and want to avoid context leakage between unrelated coding tasks.

No specific dependencies or components are required to start. You can use the provided template prompts to break a plan into per-task work and immediately begin testing the multi-agent workflow in your environment.

The best way to coordinate subagents is using a central controller with template prompts. This orchestrates fresh implementer subagents for each task and sequentially routes their output to spec and code-quality reviewers for fast, compliant results.