Skip to main content

Workflows and execution

This section covers advanced implementation details for agent-backed scopes. You’ll learn how the nullplatform agent communicates with your runner, how to define workflows for your scope’s actions, and how to debug or override them.

When should you use this?​

This content is especially useful if:

  • You're creating a custom runner for your agent-backed scope
  • You want full control over the execution behavior of actions like create, start_blue_green, or delete
  • You want to merge, skip, or replace steps in your workflow dynamically

Scope repository examples​

See the Scopes repository for production‑oriented templates you can adopt and extend. The repository follows a layout similar to the one below to implement agent-backed scopes.

Typical layout​

scopes/
├── k8s/                        # Kubernetes-specific implementation
│   ├── scope/                  # Scope lifecycle actions
│   │   └── workflows/
│   │       ├── create.yaml
│   │       ├── update.yaml
│   │       └── delete.yaml
│   ├── deployment/             # Deployment strategies (initial, blue_green, ...)
│   │   └── workflows/
│   │       ├── initial.yaml
│   │       ├── blue_green.yaml
│   │       └── ...
│   └── ...
├── service                     # Routes actions based on NP_ACTION_CONTEXT
├── agent/                      # Agent deployment scripts
├── configure                   # Configuration script
└── entrypoint                  # Main execution script

What's next?​

Looking for step-by-step examples or deeper customization?