workflow-runner

To run multi-role YAML workflows inside Claude Code or Cursor, you provide the YAML file directly in your current session. The tool uses the active LLM as the execution engine to resolve roles, render task templates, and execute steps sequentially or in parallel.

Yes, you can execute agency-orchestrator workflows without an API key by leveraging the current session's LLM directly within supported AI tools. This approach uses the active session context to handle task decomposition and workflow execution without external authentication.

YAML-based workflow orchestration processes steps sequentially or in parallel based on the workflow metadata. It reads the YAML structure, resolves role directories, renders task templates, executes the defined steps, and stores the outputs for downstream consumption.

Yes, the workflow runner supports executing steps in order or parallel based on the YAML configuration. It stores the output generated from each executed step, making it available for downstream steps to consume during multi-role collaboration.

The best way to trigger multi-role collaboration in an AI coding assistant is to provide a YAML workflow file during your session. The tool reads the workflow metadata to orchestrate and execute the defined roles and tasks automatically.

Using the current LLM session for YAML workflows eliminates the need for separate API keys and external automation setups. It directly leverages the active AI tool environment to handle task decomposition, step execution, and output storage seamlessly.