workflow-runner

Run multi-role YAML workflows directly in your current AI session by providing a workflow file, which the runner parses to resolve roles, handle dependencies, and execute steps in sequence or parallel without external API keys.

Yes, you can use it in Cursor for task decomposition and planning by executing structured workflow steps where the model adopts different roles to handle dependencies and render template variables.

Multi-agent orchestration via YAML is executing structured workflow definitions inside an AI session where the model sequentially adopts different roles, resolves input variables, and handles dependency layers for deterministic task execution.

No, you can execute multi-role collaboration workflows directly inside your AI chat session without setting up separate automation infrastructure, as the runner handles dependency resolution and output persistence internally.

It resolves template variables and step dependencies by parsing the YAML definition, extracting inputs, building dependency layers, and rendering outputs to execute each step with the correct role context and failure handling.

While parallel step execution is supported, limitations depend on the AI session's context window capacity and the complexity of dependency layers, though the runner persists outputs and handles failures to maintain workflow determinism.