What problem does it solve? Running multi-agent YAML workflows normally requires a CLI runtime and API keys. This Skill lets you execute agency-orchestrator workflows directly inside Claude Code, OpenClaw, or Cursor, using the current conversation's LLM to play each role. ## Core Features & Use Cases - YAML Workflow Parsing: Reads workflow files, extracts inputs, steps, roles, and dependency graphs, then builds a layered topological execution plan. - Multi-Role Orchestration: Loads agent role definitions from agency-agents-zh, renders task templates with variables, and executes steps sequentially or in parallel via sub-agents. - Structured Output Persistence: Saves each step's output, a final summary, and metadata.json into a dated .ao-output directory. - Use Case: A user says "run workflows/story-creation.yaml" or "have a product manager and architect review this PRD" — the Skill parses or generates the workflow, collects required inputs, and executes each role step by step. ## Quick Start Run the workflow defined in workflows/story-creation.yaml using the roles from agency-agents-zh and save the results.
