Network Node
What Network is and why it exists
Network is the composability node. It allows one workflow to call another workflow as a sub-network. This is how NodeFox supports modular architecture without duplicating logic across dozens of graphs.
In practice, Network is the backbone for scalable workflow programs. Teams can package stable patterns once, then reuse them across products, departments, and environments with controlled inputs and outputs.
Execution behavior
When a Network node runs, parent input values are mapped into child-network variables, the child network executes, and configured child outputs are returned to parent output slots. This contract-based exchange keeps parent and child responsibilities clear.
Because Network calls are explicit in the graph, they are easier to review than hidden function calls scattered across scripts.
In Network View, you can inspect a child network quickly: hold Option and click a Network node to open its sub-network.
Configuration details
| Field | What it controls | Practical guidance |
|---|---|---|
Network Name | Target child network to invoke | Use versioned naming conventions so parents do not silently drift to unintended child behavior. |
Input Variable Names | Parent slot -> child variable mapping | Keep mappings documented and stable across consuming workflows. |
Output Variable Names | Child variable -> parent output mapping | Keep output contracts narrow and typed to prevent downstream ambiguity. |
| Activation and fallback routing | Parent behavior on child completion/failure | Define explicit parent-side handling for child timeout, validation failure, and partial result cases. |
A good contract is small, explicit, and version-aware.
Modular architecture patterns
A common pattern is layered pipelines such as Network(preprocess) -> Network(classify) -> Network(format), where each child network has a clear single responsibility. Another pattern is shared governance modules, where multiple parent workflows call one policy-evaluation network before any high-impact write path.
Network is also useful for internal platform teams that publish reusable building blocks for business teams. This keeps architecture consistent while allowing domain-specific variation in parent workflows.
High-value module categories:
- utility modules: shared normalization/validation stages
- policy modules: approval and governance gates
- domain modules: reusable business logic pipelines
- integration modules: standardized adapter behavior for external systems
Versioning and promotion strategy
Treat child networks like reusable software modules:
- pin parent workflows to known child versions
- test new child versions in staging with replay inputs
- promote with compatibility notes and contract diff review
- monitor incident impact radius after shared-module updates
This minimizes surprise breakage when one child network is reused across many parent workflows.
Common mistakes to avoid
The most common issue is unstable contracts. If child input/output names change without coordination, downstream parent behavior can fail in subtle ways. Versioning and contract docs are essential.
Another mistake is over-nesting sub-networks without clear ownership. Deep nesting can make debugging harder unless run evidence and ownership boundaries are clear.
Operational signals
- child-network invocation latency and variance
- contract mismatch failures (missing input/output variable names)
- parent fallback rate from child errors
- rollout defect rate after child-network version upgrades