Exercise 3 — Memory on disk and a human approval step

Time

35 min

You need

the kit, the venv, Ollama running, week12/diag_tools.py and week12/exercise_2_solution.py in place

Deliverable

the output of the four runs, and the number of rows in the checkpoints table

The lab kit of the course: https://github.com/hrhouma2/aiopsatlas-ml-data-diagnostics-labs-en

Goal

Two changes to the agent of Exercise 2. First, a checkpointer: the state is saved in data/checkpoints.sqlite after every node, under a thread_id, so a technician can ask a follow-up question and the agent remembers. Second, a gate: before any SQL that reads the incidents table, the graph pauses with interrupt() and shows the query to a human. The human says yes or no, and the graph resumes. You will run the same thread twice to prove the memory, and a new thread to prove the forgetting.