Part 1: The Mental Shift · Module 4
Your First Agent: A Loop, a Tool, and a Stop Condition
Ready. Lessons, visuals, quiz, harness and check script are complete and were run on a current model.
Work through it in order
- 1 The loop, drawn onceSix questions went through one two-tool agent and took different numbers of model calls, from a single call to several, with no change to the code between them. 1 interactive
- 2 Two exits and a capThe step cap never fired on the six questions, the longest of them came within one step of it, and the harness that produced the table would have hidden a cap hit because it had no raise. 1 interactive
- 3 Steps you did not writeThe model split two dependent tool calls across steps, sent two independent ones together in a single response, retried a failed lookup with a different word, used the calculator for bare arithmetic and called nothing for a general question. 2 interactives
- 4 Where the loop breaksReading only the first content block drops a second tool request, in the worst case without any error, answering one of two requests was rejected by the API, a failed tool did not stop the model, and the same six prompts run twice gave different traces. 2 interactives
- Q QuizEvery option explained, scored as you go.
- T Hands-on testThe check script passes on your trace and your prediction: every model call has a line with its stop reason, tool requests and usage, all six questions end on end_turn, the forced run ends at a cap of one with a raise, and the prediction is dated before the first call. Then you tick the three checklist lines the script cannot see, because it reads your trace and not your code.