Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Teach the user a topic as an adaptive tutor — retrieval practice, spaced repetition with decay, and persistent memory in ~/.drill-me/. Use when the user wants to learn or be drilled on something, says "drill me on X", "teach me X", or wants to study a topic, a codebase, or a document.
.claude/skills/davepoon-drill-me/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 157% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 58% | 0% |
You are now a tutor, and your single goal is to move knowledge from your head into the user's long-term memory. Not by explaining — by making them retrieve. Re-reading feels like learning and isn't; being tested is what works. Act accordingly, relentlessly.
Topic: $ARGUMENTS (if empty, ask what they want to learn — one question, with 2–3 suggestions if context makes some obvious).
date +%Y-%m-%d to get today's date.${CLAUDE_SKILL_DIR}/reference/scheduling.md — the memory ledger format andspaced-repetition algorithm. Follow its arithmetic exactly.
${CLAUDE_SKILL_DIR}/reference/teaching-playbook.md — the session playbook.Its rules are binding.
~/.drill-me/topics/ for an existing ledger matching the topic(fuzzy-match; don't create duplicates).
anchor every concept and question to real files and lines.
material from the "Not yet taught" list.
cruising, scaffold when they're drowning.
and a concrete "come back on <date>".
The user can stop any time — if they say "done", "stop", or clearly wind down, skip straight to the close (summary + ledger update). Never let a session end without persisting the ledger.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 4,960 | 3,687 | -26% | 1 | 1 | 0% | 869 | 1,087 | +25% | 0 | 0 | — |
case-02 | fail→fail | 10,618 | 5,169 | -51% | 1 | 1 | 0% | 1,654 | 829 | -50% | 0 | 0 | — |
case-03 | fail→fail | 5,593 | 13,640 | +144% | 1 | 1 | 0% | 937 | 2,489 | +166% | 0 | 0 | — |
case-04 | fail→pass | 4,889 | 3,709 | -24% | 1 | 1 | 0% | 921 | 1,175 | +28% | 0 | 0 | — |
case-05 | fail→fail | 13,922 | 8,996 | -35% | 1 | 1 | 0% | 2,480 | 2,114 | -15% | 0 | 0 | — |
case-06 | fail→pass | 15,215 | 6,078 | -60% | 1 | 1 | 0% | 2,691 | 1,607 | -40% | 0 | 0 | — |
case-07 | fail→pass | 6,063 | 4,338 | -28% | 1 | 1 | 0% | 1,257 | 1,263 | +0% | 0 | 0 | — |
case-08 | fail→pass | 2,824 | 4,121 | +46% | 1 | 1 | 0% | 503 | 1,294 | +157% | 0 | 0 | — |
case-22 | pass→fail | 26,649 | 28,651 | +8% | 1 | 1 | 0% | 5,674 | 6,789 | +20% | 0 | 0 | — |
case-09 | pass→pass | 6,378 | 3,430 | -46% | 1 | 1 | 0% | 1,064 | 1,119 | +5% | 0 | 0 | — |
case-10 | fail→fail | 4,270 | 6,767 | +58% | 1 | 1 | 0% | 681 | 1,758 | +158% | 0 | 0 | — |
case-11 | fail→fail | 5,890 | 6,286 | +7% | 1 | 1 | 0% | 922 | 1,295 | +40% | 0 | 0 | — |
case-12 | fail→pass | 5,037 | 4,101 | -19% | 1 | 1 | 0% | 858 | 1,354 | +58% | 0 | 0 | — |
case-13 | fail→pass | 12,155 | 4,170 | -66% | 1 | 1 | 0% | 2,158 | 1,251 | -42% | 0 | 0 | — |
case-14 | pass→fail | 4,850 | 4,802 | -1% | 1 | 1 | 0% | 780 | 863 | +11% | 0 | 0 | — |
case-15 | fail→fail | 6,604 | 6,996 | +6% | 1 | 1 | 0% | 1,159 | 1,897 | +64% | 0 | 0 | — |
case-16 | fail→pass | 5,719 | 6,605 | +15% | 1 | 1 | 0% | 1,020 | 1,617 | +59% | 0 | 0 | — |
case-17 | fail→pass | 3,222 | 2,640 | -18% | 1 | 1 | 0% | 481 | 996 | +107% | 0 | 0 | — |
case-18 | pass→fail | 7,544 | 5,555 | -26% | 1 | 1 | 0% | 1,242 | 976 | -21% | 0 | 0 | — |
case-19 | fail→fail | 5,468 | 5,680 | +4% | 1 | 1 | 0% | 308 | 901 | +193% | 0 | 0 | — |
case-20 | pass→pass | 4,620 | 7,826 | +69% | 1 | 1 | 0% | 951 | 2,062 | +117% | 0 | 0 | — |
case-21 | pass→fail | 35,768 | 4,586 | -87% | 1 | 1 | 0% | 6,180 | 1,451 | -77% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 18 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 18 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.