Capability nodes
Select a capability and an execution route that can run on your device.
A capability identifies the action, such as document.read, knowledge.search or image.generate. In the node editor, choose the capability and confirm the available connection, target or model for this workflow. A capability can appear in an imported workflow before its executor or local binding is ready.
- Check setup issues on the node. A connection, knowledge-base binding, browser permission or media model can be required depending on the capability.
- Map the required inputs from the run or an upstream node. Use a small run to inspect the executor’s actual output.
- If two actions have different permissions or results, give each an explicit step. This makes it easier to identify which action ran, which result reached downstream nodes and which policy stopped execution.