subagent-driven-development

You orchestrate tasks by dispatching a dedicated subagent for each task and applying a strict two-stage review process. This approach ensures fresh context boundaries and enables fast, isolated, high-quality iterations within a single session.

The two-stage review process first checks spec compliance to catch gaps early, then evaluates code quality to improve reliability. This structured workflow ensures that the dispatched subagent output meets all requirements before completing the task.

You standardize workflows by using ready-made prompt templates for the implementer, spec reviewer, and code-quality reviewer. These templates provide explicit prompts and per-task context to ensure spec compliance and code quality.

Yes, you can run in-session, parallel-safe execution for mostly independent tasks. The workflow provides clear handoffs and automatic progress tracking to maintain isolated context while enabling rapid iteration within a single session.

You should not use subagent-driven development if your tasks are heavily dependent on each other or require multi-session execution. It targets interactive sessions where tasks are mostly independent and execution stays within a single session.

Yes, subagent-driven development requires explicit prompts, per-task context, implementer and reviewer templates, and a controlled workflow. These inputs ensure spec compliance and code quality during rapid agile iterations.