I’m testing whether DeepSeek can handle a workflow where it calls one tool, reads the result, and then decides what to do next. This is different from a single lookup: the second action may depend on details returned by the first, and sometimes the right move is to stop rather than call anything else.

Has anyone tried this with a real application flow? I’m curious about how reliably DeepSeek chooses the next tool and passes along the relevant parts of the earlier result, particularly when the first response is incomplete or unexpected.