Separating a capability from its connector.
An operator defines the work it needs. A connector binds that work to a particular system.
Fibric’s operator contract names capabilities rather than embedding a vendor integration in the task logic. A deployment binds those capabilities to connectors for its source systems. This is how a platform can support different customers without putting every vendor’s API inside every operator.
For example, an operator that needs order context can depend on a commerce capability. The connector translates between that contract and the configured commerce system. Authentication, field mapping, pagination, and error handling belong to the integration.
Changing a system.
Swapping a connector is a configuration change only after compatibility is established. Two systems may represent status, permissions, and partial completion differently. A replacement needs to satisfy the capability’s data and behavior requirements; matching the operation name is not enough.
The same principle applies to equipment. A protocol can expose a point or command without proving that it has the meaning, limits, or authority the operator expects. Those conditions require deployment-specific validation.
The marketplace gives each listing a declared role, requirements, inputs, and proposed actions. Those details help identify a candidate integration. They are the starting point for review, not an installation guarantee.