Label-Driven Prompt Generation Architecture When building issue-to-prompt automation, use Python + gh CLI for structured classification: parse issue labels (e.g., ) to map to prompt templates, extract plan files or metadata from issue bodies using path tables, and implement both single-issue and batch-query modes. Store scripts in , mark transient output directories in , and verify classification against real issues before batch deployment. Test label matching, plan extraction, and batch filtering in sequence to catch routing logic errors early. ---