When an agent invokes a Hatz Workshop workflow, it appears unable to reliably pass collected user answers into the workflow’s actual input variables, even when the workflow schema is known.

There also seems to be a second limitation: after the workflow completes, the agent cannot reliably consume, inspect, or display the workflow’s resulting outputs back to the user (for example, generated documents, context packs, or structured results).

Requested improvement

Please enable:

Explicit input binding from agent-collected values to workflow input fields

Reliable workflow execution with provided inputs

Agent access to workflow outputs after completion

Output rendering/attachment support so the agent can present generated files, summaries, and context directly in chat

This would make agents much more useful as a front door for workflows, especially for intake-driven use cases which require more complex workflows to complete.