Your Claude Code usage
Demo account. Everyone and everything here is made up: a class of invented students, analyzed by the same code as a real one. Nothing you change is saved. 84 typed prompts in 8 sessions, and 7.6 hours of active time, over 3 assignments. Your agent-ability portrait: why it looks like this ↓ In class we went over common pitfalls in working with a coding agent. The course staff checked your Claude Code sessions in each assignment for each of them. “Yes” means the record shows it happened; “Likely” means the pattern is there but not what it cost. Below the table, each one that happened, with what you typed, what happened, and what to do next time. Assignment 1: Two fixes did not hold. You asked for the streaming fix and the tree fix again after the agent had finished each one. Before you move on from a fix, ask for evidence that it works. Assignment 2: You asked for the run form again and again in one session, each time after the agent said it was done. Ask it to show the form working firs