Subagent Task Execution

If you're working through a multi-step plan or task breakdown, this skill gives you a structured way to execute it without cutting corners. It enforces a one task, one subagent approach with mandatory review stages: you prep a todo list, dispatch a specialized subagent for each task with skill discovery beforehand, then run both spec compliance and quality reviews before marking anything complete. The workflow is rigid by design, with explicit red flags like skipping reviews or batching tasks together. It's overkill for simple work, but if you need guardrails to prevent sloppy execution when juggling complex dependencies, the structured handoffs and verification loops actually enforce the discipline most people skip when they're moving fast.

Install to Claude Code

npx -y skills add davidkiss/smart-ai-skills --skill subagent-task-execution --agent claude-code