dag-executor

dag-executor is an orchestration skill that decomposes complex natural language tasks into parallelizable agent graphs. It intelligently maps subtasks to available skills and executes them concurrently using Claude Code's Task tool. Use this when you need to automate and parallelize multi-step workflows through AI-driven task decomposition.

This skill analyzes DAG execution history to identify successful workflow patterns and detect anti-patterns, providing optimization recommendations. Use it when prompted to "learn patterns" or "optimize based on history" to improve future DAG performance. It pairs with execution tracers and graph builders to apply learned insights.

This skill monitors iteration progress by tracking quality trends and detecting plateaus to recommend optimal stopping points. Use it when you need to determine when further iterations won't improve task outcomes. It pairs with other DAG skills like the iteration detector and feedback synthesizer for comprehensive workflow monitoring.

This skill executes scheduled DAG workflows in parallel waves using the Task tool, managing concurrent agent spawning within set resource limits. It coordinates task execution and agent handoffs for workflow runs when triggered by commands like 'execute dag' or 'run workflow'. Use this specifically for parallel execution, not for building DAGs or scheduling them.