workflow-runner

To run multi-role YAML workflows inside an AI tool, provide a YAML workflow file to the session. The system parses inputs, resolves dependencies, and executes topological sequencing to manage multi-role collaboration without external APIs.

Yes, you can execute agency-orchestrator workflows directly in Claude Code, OpenClaw, or Cursor without API keys. The execution handles role-based steps, variable substitution, and in-session dependency resolution natively.

Topology-driven execution for multi-role AI collaboration is a process that parses workflow inputs, resolves step dependencies, renders role prompts, and produces step-level outputs for reuse in later steps within a single session.

You can orchestrate product planning and architecture reviews within a single AI session by supplying a YAML workflow. The system enforces per-step role emulation and stores results to produce a final summary.

Yes, workflow execution supports variable substitution from both initial inputs and prior step outputs. It renders role prompts dynamically and stores intermediate results to ensure correct topological sequencing across the workflow.

You should not use in-session YAML workflow orchestration if your multi-role tasks require external API integrations or stateful execution across multiple disconnected sessions, as this approach is designed for self-contained, single-session dependency resolution.