subagent-driven-development

Subagent-driven development deploys a fresh subagent for each discrete task within your session, ensuring isolated reasoning to prevent context leakage. This approach enables parallel work where appropriate while maintaining structured task dispatch.

A two-stage review workflow first checks spec compliance, then evaluates code quality. This pipeline enforces correctness by having dedicated reviewer templates assess implementation against specifications before reviewing the actual code standards.

Break complex work into independent tasks by using per-task prompt flows and frontmatter metadata to assign each task to a dedicated subagent. This structured approach isolates reasoning and enforces review before integration.

Yes, subagent-driven development supports parallel task dispatch within a single session. By deploying independent subagents, you can handle mostly independent tasks concurrently while managing completion statuses and orchestrating reviewers.

Subagent orchestration manages blockers through safe escalation paths. When insufficient context occurs or blockers arise, the workflow reports task statuses and escalates issues to ensure proper resolution before integration.

Use a subagent-driven development workflow when tasks are mostly independent and require isolated reasoning. It is ideal for complex projects needing a structured two-stage review process to ensure correctness and speed before integration.