Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fix slides that are documents in landscape mode — the one-point-per-slide rule, the projection-vs-reading fork that decides density, the text diet (headlines and evidence, prose to notes), and the glance test. Use when asked my slides are too busy, how much text per slide, fix this wall-of-bullets deck, or make this readable from the back of the room. Produces the density diagnosis, the per-slide fixes (split, strip, or move-to-notes), the projection/document fork decision, and the glance-test r
.claude/skills/mohitagw15856-slide-density-rules/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 52% | 0% |
The wall-of-bullets slide fails twice simultaneously: the audience reads instead of listening (reading speed beats talking speed, so they finish and disengage while the presenter catches up), and nothing on it is retained because everything competed. The fix starts with a fork most decks never take: is this projected (accompanying a live talk — then slides are billboards: one point, big, glanceable) or read (sent as a document — then it's a doc wearing slide clothes, and density is fine but it should be built as a outline-before-prose memo-deck on purpose)? Hybrid decks that try both do both badly; the rules below are the fork, then the diet.
Ask for these if not provided:
The mode with its reasoning · if both-needed: the two-artifact plan]
| Slide | Points on it | Glance verdict | Fix | |---|---|---|---| Split / strip-to-spoken / demote-to-notes / appendix]
What moved where: notes ← prose · appendix ← detail · leave-behind ← the full version]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 29,921 | 19,728 | -34% | 1 | 1 | 0% | 3,338 | 3,535 | +6% | 0 | 0 | — |
case-02 | fail→pass | 29,703 | 22,508 | -24% | 1 | 1 | 0% | 3,669 | 4,657 | +27% | 0 | 0 | — |
case-03 | fail→pass | 23,723 | 27,581 | +16% | 1 | 1 | 0% | 2,435 | 4,235 | +74% | 0 | 0 | — |
case-04 | pass→pass | 24,495 | 27,331 | +12% | 1 | 1 | 0% | 2,464 | 4,433 | +80% | 0 | 0 | — |
case-05 | pass→pass | 21,459 | 25,586 | +19% | 1 | 1 | 0% | 3,060 | 4,238 | +38% | 0 | 0 | — |
case-06 | pass→pass | 21,924 | 21,363 | -3% | 1 | 1 | 0% | 2,223 | 3,887 | +75% | 0 | 0 | — |
case-07 | fail→pass | 19,400 | 19,840 | +2% | 1 | 1 | 0% | 2,054 | 3,195 | +56% | 0 | 0 | — |
case-08 | fail→pass | 19,282 | 17,262 | -10% | 1 | 1 | 0% | 2,151 | 3,001 | +40% | 0 | 0 | — |
case-09 | fail→pass | 16,623 | 16,951 | +2% | 1 | 1 | 0% | 1,768 | 2,693 | +52% | 0 | 0 | — |
case-10 | fail→pass | 18,149 | 18,155 | +0% | 1 | 1 | 0% | 1,960 | 3,064 | +56% | 0 | 0 | — |
case-11 | pass→pass | 19,377 | 18,025 | -7% | 1 | 1 | 0% | 1,957 | 3,056 | +56% | 0 | 0 | — |
case-12 | pass→pass | 16,216 | 24,400 | +50% | 1 | 1 | 0% | 2,219 | 3,649 | +64% | 0 | 0 | — |
case-13 | pass→pass | 20,942 | 17,363 | -17% | 1 | 1 | 0% | 2,232 | 3,244 | +45% | 0 | 0 | — |
case-14 | fail→pass | 24,825 | 22,074 | -11% | 1 | 1 | 0% | 2,604 | 3,196 | +23% | 0 | 0 | — |
case-15 | fail→pass | 18,393 | 21,050 | +14% | 1 | 1 | 0% | 1,922 | 3,315 | +72% | 0 | 0 | — |
case-16 | pass→pass | 20,119 | 21,206 | +5% | 1 | 1 | 0% | 2,020 | 3,263 | +62% | 0 | 0 | — |
case-17 | pass→pass | 19,037 | 21,005 | +10% | 1 | 1 | 0% | 1,835 | 3,055 | +66% | 0 | 0 | — |
case-18 | pass→pass | 13,249 | 8,969 | -32% | 1 | 1 | 0% | 1,161 | 1,750 | +51% | 0 | 0 | — |
case-19 | pass→pass | 16,458 | 9,695 | -41% | 1 | 1 | 0% | 1,712 | 2,487 | +45% | 0 | 0 | — |
case-20 | pass→pass | 21,255 | 16,371 | -23% | 1 | 1 | 0% | 1,892 | 2,610 | +38% | 0 | 0 | — |
case-21 | fail→pass | 10,847 | 12,057 | +11% | 1 | 1 | 0% | 1,499 | 2,188 | +46% | 0 | 0 | — |
case-22 | fail→pass | 18,738 | 20,388 | +9% | 1 | 1 | 0% | 2,122 | 3,398 | +60% | 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 +45 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.