agentleFS
jaskirat1616jaskirat1616/grok-skills/skills/presentation-builder/SKILL.md

presentation-builder

Skill22 starsChanged 3 months agoMITView on GitHub

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".

14 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    USE for: structuring a new deck, converting a document/report into slides, reordering a rambling deck, writing speaker notes, sharpening a …

    line 23
  2. Distinctive:

    DON'T over-engineer for: a single status slide, or when the user already has a finished deck and only wants visual styling (that's a design …

    line 24
  3. Distinctive:

    Follow these steps in order.

    line 28
  4. Distinctive:

    Before writing any slide, establish (ask the user if unknown, max 3-4 crisp questions):

    line 32
  5. Distinctive:

    If the user gave a document, infer drafts of these from it, then confirm the core message before proceeding.

    line 41
  6. Very specific:

    Write the core message as ONE declarative sentence (see references/storytelling-frameworks.md → "Core Message Test").

    line 45
  7. Very specific:

    Every slide gets the fields in templates/slide-outline.md:

    line 74
  8. Very specific:

    Run every slide and the whole deck through references/deck-review-checklist.md.

    line 83
  9. Distinctive:

    Write speaker notes as speech, not bullet points — read them aloud to test.

    line 102
  10. Distinctive:

    Use the 1-7-7 sanity guide loosely (≤7 lines, ≤7 words) but prefer even less.

    line 103

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Presentation Builder

Overview

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:

  1. One core message — the whole deck exists to make ONE point land.
  2. A narrative arc — slides progress like a story (tension → resolution), not a list of facts.
  3. One idea per slide — each slide advances the argument by exactly one step, with spoken-word speaker notes.

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.

When to use vs. not use

  • USE for: structuring a new deck, converting a document/report into slides, reordering a rambling deck, writing speaker notes, sharpening a pitch.
  • DON'T over-engineer for: a single status slide, or when the user already has a finished deck and only wants visual styling (that's a design task, not a structure task).

Workflow

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.

Step 1 — Discover the four anchors

Before writing any slide, establish (ask the user if unknown, max 3-4 crisp questions):

AnchorQuestion to answer
AudienceWho is in the room? What do they already know, care about, and fear?
Goal / AskWhat should they DO or DECIDE after the talk? (One verb.)
Core messageThe single sentence they should remember tomorrow.
ConstraintsTime 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.

Step 2 — Lock the core message

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.

Step 3 — Choose a narrative arc

Pick the arc that fits the goal. Defaults:

  • Pitch / persuade / sell → "What is — What could be" (problem/possibility oscillation, ending on a call to action).
  • Inform / teach → "Pyramid" (answer first, then supporting groups of ideas).
  • Report / update → "Situation → Complication → Resolution → Next steps."
  • Inspire / keynote → "Hero's journey lite" (status quo → catalyst → struggle → transformation).

See references/storytelling-frameworks.md for full descriptions and when to use each.

Step 4 — Draft the slide spine

Lay out section headers first (3-5 sections for a typical talk), then slides under each. Use the canonical spine:

  1. Hook (1 slide) — a question, surprising stat, or vivid scene. No agenda slide as slide 1.
  2. Stakes (1-2) — why this matters now, to THIS audience.
  3. Core message (1) — state it plainly, early.
  4. Body (3-7) — the argument, one idea per slide, each building on the last.
  5. Proof (1-3) — evidence, demo, case study, data.
  6. Call to action (1) — the specific ask / next step.
  7. Close (1) — restate core message; memorable final line.

Adjust counts to the time budget (rule of thumb: ~1-2 minutes per content slide; see scripts/pacing.py).

Step 5 — Write each slide using the slide contract

Every slide gets the fields in templates/slide-outline.md:

  • Slide title — a full assertion ("Sales fell because onboarding broke"), NOT a topic label ("Sales").
  • One idea — the single point.
  • On-slide content — minimal: a phrase, one chart, or one image cue. No paragraphs.
  • Speaker notes — what you actually SAY (conversational, 30-90 sec worth).
  • Transition — one line linking to the next slide ("But there's a catch...").

Step 6 — Self-review with the checklist

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.

Step 7 — Deliver

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.

Decision Heuristics

  • Assertion titles — if a title could be a slide in any deck, it's too vague. Make it say something.
  • The "so what?" test — after each slide, the audience should never think "so what?". The transition answers it.
  • Cut to clarify — when unsure whether to include a slide, cut it. Fewer slides, more impact.
  • Data needs a verdict — never show a chart without stating what it proves on the slide title.
  • Demos > descriptions — if you can show it, show it; replace a "features" slide with a proof slide.
  • One emotion per section — curiosity, concern, hope, urgency. Map the emotional beat.

Best Practices

  • Open with tension, not logistics. Save the agenda (if any) for after the hook.
  • State the core message in the first 90 seconds AND at the end. Repetition is the medium.
  • Write speaker notes as speech, not bullet points — read them aloud to test.
  • Use the 1-7-7 sanity guide loosely (≤7 lines, ≤7 words) but prefer even less; aim for one phrase.
  • Each chart answers one question. Title the chart with the answer.
  • End on a sentence, not a "Thank you / Questions?" slide. Land the message, then thank.

Common Pitfalls

  • Topic titles ("Background", "Results") instead of assertions. Always rewrite.
  • The everything slide — multiple ideas crammed together. Split it.
  • Reading the slide — slide text duplicates speech. Slide = visual aid; notes = words.
  • Burying the ask — call to action vague or missing. Make it one concrete verb + deadline.
  • No arc — a flat list of facts. Add tension/resolution beats.
  • Death by data — slide after slide of charts with no narrative thread.
  • Skipping audience analysis — generic deck that lands with no one.

Bundled Files

  • 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.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.