plan-and-execute-by-subagent

Automate large-scale code generation by delegating work to subagents in batches of five tasks, maintaining clean context boundaries to prevent context drift and ensure output consistency. This approach preserves context quality during distributed task execution.

Generate models, services, configs, and tests by extracting design documents into atomic tasks and orchestrating them through subagents. The process manages dependencies and tracks progress in a detailed PLAN.md file with task statuses.

Yes, you can use subagents to batch code generation tasks across different languages and stacks. The workflow supports design-driven code generation across multiple modules, handling batch creation of files from architecture documents.

Planning code generation with subagents works by extracting design documents into atomic tasks, spawning a fresh subagent every five tasks, and outputting a traceable PLAN.md that manages dependencies and tracks task statuses throughout the workflow.

Yes, you need one or more design documents to start distributed code generation. The skill extracts these documents into atomic tasks, plans the execution, and spawns subagents in batches of five to produce the code files.