Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Outlines compelling presentations built around a single core message, a clear narrative arc, one idea per slide, and concrete speaker notes. Use this skill when the user wants to create, outline, structure, or improve a slide deck, talk, pitch, keynote, conference presentation, sales demo, lecture, or any "build me slides / a presentation / a pitch deck" request — including requests like "turn this doc into slides", "help me prepare a talk", "make a deck about X", or "tighten my presentation".
.claude/skills/jayrha-presentation-builder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 93% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 138% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 43% | 0% |
This skill turns a topic, document, or rough idea into a tight, persuasive presentation outline. It enforces three rules that separate good decks from forgettable ones:
Keywords: presentation, slide deck, slides, pitch deck, keynote, talk, lecture, demo, storytelling, narrative arc, speaker notes, outline, slide outline, deck structure, conference talk, sales pitch.
The output is a structured outline (slide-by-slide), NOT a visually rendered deck. Hand the outline to PowerPoint, Google Slides, Keynote, or a tool like Marp/reveal.js.
Follow these steps in order. Do not skip the discovery step — a deck built without knowing the audience and the ask is the #1 cause of bad presentations.
Before writing any slide, establish (ask the user if unknown, max 3-4 crisp questions):
| Anchor | Question to answer | |--------|-------------------| | Audience | Who is in the room? What do they already know, care about, and fear? | | Goal / Ask | What should they DO or DECIDE after the talk? (One verb.) | | Core message | The single sentence they should remember tomorrow. | | Constraints | Time limit, slide count, format (live/async), tone. |
If the user gave a document, infer drafts of these from it, then confirm the core message before proceeding.
Write the core message as ONE declarative sentence (see references/storytelling-frameworks.md → "Core Message Test"). Every slide must earn its place by supporting this sentence. If a slide doesn't, cut it.
Pick the arc that fits the goal. Defaults:
See references/storytelling-frameworks.md for full descriptions and when to use each.
Lay out section headers first (3-5 sections for a typical talk), then slides under each. Use the canonical spine:
Adjust counts to the time budget (rule of thumb: ~1-2 minutes per content slide; see scripts/pacing.py).
Every slide gets the fields in templates/slide-outline.md:
Run every slide and the whole deck through references/deck-review-checklist.md. Fix violations. Most common: topic-label titles, multi-idea slides, missing call to action, no transitions.
Output the full outline using templates/slide-outline.md structure. Offer to: (a) generate Marp/reveal.js markdown, (b) tighten to a shorter time, or (c) write a 30-second elevator version of the core message.
references/storytelling-frameworks.md — narrative arcs, the core-message test, emotional mapping, opening/closing patterns.references/deck-review-checklist.md — per-slide and whole-deck review checklist.templates/slide-outline.md — the fill-in outline format for the final deliverable.examples/saas-pitch-deck.md — a complete worked example (input brief → finished outline).scripts/pacing.py — computes slide budget and flags over/under-packed decks from time + slide count.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | 19,549 | 22,034 | +13% | 1 | 1 | 0% | 3,500 | 5,525 | +58% | 0 | 0 | — |
case-01 | fail→fail | 21,312 | 22,252 | +4% | 1 | 1 | 0% | 3,865 | 5,119 | +32% | 0 | 0 | — |
case-02 | fail→fail | 25,884 | 21,599 | -17% | 1 | 1 | 0% | 4,548 | 5,607 | +23% | 0 | 0 | — |
case-04 | pass→pass | 13,533 | 10,299 | -24% | 1 | 1 | 0% | 2,159 | 3,422 | +58% | 0 | 0 | — |
case-05 | pass→pass | 9,047 | 9,173 | +1% | 1 | 1 | 0% | 1,499 | 3,127 | +109% | 0 | 0 | — |
case-06 | pass→pass | 11,713 | 34,033 | +191% | 1 | 1 | 0% | 1,877 | 2,971 | +58% | 0 | 0 | — |
case-07 | fail→pass | 18,742 | 23,945 | +28% | 1 | 1 | 0% | 2,941 | 5,664 | +93% | 0 | 0 | — |
case-08 | fail→pass | 12,681 | 17,708 | +40% | 1 | 1 | 0% | 2,094 | 4,976 | +138% | 0 | 0 | — |
case-09 | fail→fail | 16,763 | 22,508 | +34% | 1 | 1 | 0% | 2,829 | 3,886 | +37% | 0 | 0 | — |
case-10 | pass→pass | 11,631 | 12,594 | +8% | 1 | 1 | 0% | 2,263 | 3,802 | +68% | 0 | 0 | — |
case-11 | pass→pass | 7,261 | 5,642 | -22% | 1 | 1 | 0% | 1,390 | 2,710 | +95% | 0 | 0 | — |
case-12 | pass→pass | 10,592 | 8,453 | -20% | 1 | 1 | 0% | 1,884 | 3,060 | +62% | 0 | 0 | — |
case-13 | pass→pass | 11,623 | 11,105 | -4% | 1 | 1 | 0% | 2,023 | 3,628 | +79% | 0 | 0 | — |
case-14 | fail→pass | 14,288 | 9,387 | -34% | 1 | 1 | 0% | 2,280 | 3,148 | +38% | 0 | 0 | — |
case-15 | pass→pass | 11,058 | 10,482 | -5% | 1 | 1 | 0% | 1,842 | 3,516 | +91% | 0 | 0 | — |
case-16 | pass→pass | 11,016 | 17,806 | +62% | 1 | 1 | 0% | 1,878 | 2,922 | +56% | 0 | 0 | — |
case-17 | pass→pass | 10,461 | 15,814 | +51% | 1 | 1 | 0% | 1,882 | 4,398 | +134% | 0 | 0 | — |
case-18 | pass→pass | 11,979 | 5,413 | -55% | 1 | 1 | 0% | 2,022 | 2,628 | +30% | 0 | 0 | — |
case-19 | pass→pass | 8,283 | 5,622 | -32% | 1 | 1 | 0% | 1,467 | 2,586 | +76% | 0 | 0 | — |
case-20 | fail→pass | 13,007 | 7,263 | -44% | 1 | 1 | 0% | 2,096 | 3,078 | +47% | 0 | 0 | — |
case-21 | pass→pass | 5,713 | 2,760 | -52% | 1 | 1 | 0% | 945 | 2,144 | +127% | 0 | 0 | — |
case-22 | fail→pass | 12,768 | 6,600 | -48% | 1 | 1 | 0% | 1,951 | 2,796 | +43% | 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 +23 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +33% |
Other measured skills in the registry, with their headline benchmark lift.