Part 1 of 4 · 1 min read
Overview
Agent projects fail quietly: no target, no baseline, an unfinishable feature list, and a demo that nobody can measure. A scope you can check turns the capstone into evidence about your engineering rather than a collection of features, and gives you something concrete to defend in an interview.
About the example, and what is real
Real: the schedule simulator, the cut-line function, the last-mile arithmetic, the evaluation-size calculation, the scope checker and 22 tests in code/mini-scope. Assumed: the capstone, its eleven features and hour estimates, the 0.5 and 0.25 spreads, the 20 hours per week, the 100-item floor and the 20 percent minimum disagreement used for item counts. Not real: no project was built, no user was interviewed, the three scopes are fictional, and the quality-versus-time curve rests on one anecdote.