quiz-melisted

Test whether the user's mental model still matches work that already exists — code the agent wrote, a PRD, a plan, a document tree — usually because they delegated it, skimmed it, or were away while it was built. Derives a tree from the artifact, then asks one multiple-choice question at a time: general framing first, then hard into the decision points, with distractors drawn from the alternatives actually rejected, or from the conventional choice this artifact departed from. Every wrong answer is triaged — did the agent drift from the user's intent (→ punch list), or is it a gap in the user's knowledge (→ explained)? Needs an artifact that already exists: for a subject nothing was built for use `tutor`, and for a plain re-orientation with no questions asked use `catchup`. Triggers: /quiz-me, "quiz me on what we built", "test me on this plan", "grill me on the code you wrote", "do I actually understand what we built", "am I still in sync with what you wrote", "I approved that without reading it".
mehdiforoozandeh/skills · ★ 0 · AI & Automation · score 75

Install: claude install-skill mehdiforoozandeh/skills

# quiz-me — is your model still the artifact's model? A lot got built while the user steered at a distance. Reading all of it defeats the point of delegating; assuming they're current is how drift survives. quiz-me tests the delta. The ground truth is an **artifact**, which is the only reason this works: the answer key is checkable on disk, so a wrong answer can mean the *artifact* is wrong and not the user. ## Target — what's under test - **No argument** → the work of the current session. Derive the tree and the answer key from the **current state of the files** — re-read what changed. Your recollection is not the artifact: edits get reverted, failing tests force silent rewrites, context gets compacted. The transcript supplies the *rejected alternatives*; it is never the answer key. If the session produced no durable artifact, say so — drift findings are weak without one. - **With an argument** → that target, cold. A file, a directory, a PRD, a document subtree, a diff, a branch. Nothing needs to be in context. - **Ambiguous** → ask once, in one line. Before question 1, state the scope in one line and name the two affordances: `D — not sure`, and that saying **object** at any point flags a choice the user dislikes. **Don't show the tree** — its structure telegraphs the answers. ## Build the tree first Derive it breadth-first: every top-level area, drilling only where a decision is visible. Build wider than you will ask, so you know where to go when an answer