awesome-workflows

ALWAYS read this skill before writing any Workflow tool script, before calling the Workflow() tool, or before spawning more than 3 agents. Covers the critical patterns that prevent wasted tokens and wrong model choices: pipeline() vs parallel(), Haiku for fan-out agents, never use agents just to write files (return results instead), schema output, phase organization, and resume. Triggers on: writing a workflow script, Workflow({ script: ... }), pipeline(), parallel(), agent() in a loop, "run in parallel", "fan-out", "orchestrate subagents", "multi-agent", "haiku for agents", "how do I save results from a workflow", "workflow burning tokens", "too many agents", "model for subagents". If you are about to write a workflow script or call Workflow() — stop and read this first.