Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Actually remember and use what you read — an active-reading system that beats the highlight-and-forget cycle. Use when asked how do I remember what I read, I forget books right after finishing, help me retain what I study, or take better reading notes. Produces an active-reading method (questions before, engagement during, retrieval after), a lightweight note format that captures the few ideas worth keeping, a spaced review touch, and how to actually apply what you read — turning passive consump
.claude/skills/mohitagw15856-reading-retention-system/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -9% | 0% |
Most reading evaporates — you finish a book and a month later couldn't say what it was about. Highlighting feels productive but does almost nothing. Retention comes from active reading: engaging before, during, and after, capturing the few real ideas, and using them. This sets up a lightweight system that turns what you read into knowledge you actually keep and apply.
Ask for these if not provided:
Before: hold a question/purpose]. During: engage — react, connect, argue — not passive highlight]. Capture (per chapter/piece): only the few key ideas, in your words, linked to what you know]. After: retrieve the key ideas from memory — don't re-read]. Review: a light spaced revisit]. Apply: how to actually use one idea].
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 27,211 | 17,418 | -36% | 1 | 1 | 0% | 2,972 | 2,665 | -10% | 0 | 0 | — |
case-02 | fail→fail | 39,663 | 28,907 | -27% | 1 | 1 | 0% | 4,561 | 3,644 | -20% | 0 | 0 | — |
case-03 | fail→fail | 24,763 | 17,317 | -30% | 1 | 1 | 0% | 3,442 | 3,116 | -9% | 0 | 0 | — |
case-04 | fail→fail | 20,461 | 14,381 | -30% | 1 | 1 | 0% | 2,983 | 2,913 | -2% | 0 | 0 | — |
case-05 | fail→fail | 76,807 | 22,725 | -70% | 1 | 1 | 0% | 3,109 | 2,963 | -5% | 0 | 0 | — |
case-06 | fail→fail | 24,855 | 16,951 | -32% | 1 | 1 | 0% | 3,300 | 2,921 | -11% | 0 | 0 | — |
case-07 | fail→fail | 16,943 | 15,965 | -6% | 1 | 1 | 0% | 2,633 | 3,035 | +15% | 0 | 0 | — |
case-08 | fail→fail | 18,465 | 14,336 | -22% | 1 | 1 | 0% | 2,875 | 2,515 | -13% | 0 | 0 | — |
case-09 | fail→pass | 36,036 | 22,384 | -38% | 1 | 1 | 0% | 2,941 | 2,623 | -11% | 0 | 0 | — |
case-10 | fail→fail | 16,438 | 13,323 | -19% | 1 | 1 | 0% | 2,732 | 2,646 | -3% | 0 | 0 | — |
case-11 | fail→fail | 18,984 | 14,558 | -23% | 1 | 1 | 0% | 2,588 | 2,702 | +4% | 0 | 0 | — |
case-12 | pass→pass | 19,213 | 14,181 | -26% | 1 | 1 | 0% | 2,965 | 2,641 | -11% | 0 | 0 | — |
case-13 | fail→pass | 18,599 | 13,948 | -25% | 1 | 1 | 0% | 2,952 | 2,892 | -2% | 0 | 0 | — |
case-14 | fail→fail | 16,916 | 20,575 | +22% | 1 | 1 | 0% | 2,303 | 2,938 | +28% | 0 | 0 | — |
case-15 | fail→pass | 20,229 | 13,654 | -33% | 1 | 1 | 0% | 3,092 | 2,843 | -8% | 0 | 0 | — |
case-16 | fail→fail | 18,689 | 14,487 | -22% | 1 | 1 | 0% | 2,623 | 2,993 | +14% | 0 | 0 | — |
case-17 | fail→fail | 25,877 | 13,647 | -47% | 1 | 1 | 0% | 3,116 | 2,933 | -6% | 0 | 0 | — |
case-18 | fail→pass | 16,873 | 15,912 | -6% | 1 | 1 | 0% | 2,696 | 2,905 | +8% | 0 | 0 | — |
case-19 | fail→pass | 19,347 | 12,603 | -35% | 1 | 1 | 0% | 2,856 | 2,589 | -9% | 0 | 0 | — |
case-20 | pass→fail | 35,340 | 20,447 | -42% | 1 | 1 | 0% | 6,532 | 3,502 | -46% | 0 | 0 | — |
case-21 | pass→pass | 17,691 | 24,473 | +38% | 1 | 1 | 0% | 2,860 | 4,109 | +44% | 0 | 0 | — |
case-22 | pass→pass | 20,365 | 36,894 | +81% | 1 | 1 | 0% | 2,784 | 5,333 | +92% | 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. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.