Skip to content
Content
Skill

/speech-coach

Drafts and critiques speeches, talks, keynotes, lightning talks, demo-day pitches, all-hands updates, conference presentations, wedding toasts, and any spoken-delivery piece using Glenn Kramon's 12 wowing-the-crowd rules from Stanford GSB's Winning Writing. Use when the user is

From plugin
winning-writing
1731 skills

Claude Code

$ npx -y skills add kalyvask/winning-writing --skill speech-coach --agent claude-code

SKILL.md

speech-coach.SKILL.md
name: speech-coach
description: Drafts and critiques speeches, talks, keynotes, lightning talks, demo-day pitches, all-hands updates, conference presentations, wedding toasts, and any spoken-delivery piece using Glenn Kramon's 12 wowing-the-crowd rules from Stanford GSB's Winning Writing. Use when the user is preparing a talk, presentation, keynote, panel opener, board presentation, all-hands, town hall, demo-day pitch, toast, eulogy, commencement address, or any piece that will be delivered aloud. Triggers on phrases like "I'm giving a talk," "preparing a speech," "keynote," "all-hands," "demo day pitch," "wedding toast," "panel opener," "TED-style," "deliver this aloud," "speaking at," "presentation script."

Speech coach

Source: `points/speech-rules.md` (the 12 rules + the order-of-operations drafting flow + the Sparkline), `points/delivery-rules.md` (the three Vs, derailers, Q&A structures — for Mode 3 coaching), `points/kramon-master.md` (rhythm, jargon, warmth), `points/banned-jargon.md` (especially for spoken delivery — every banned phrase hits even harder when read aloud).

What this skill does

Given a topic + audience (or an existing draft), produce a talk shaped for the **ear**, not the page. The most common bad-talk failure mode is a memo read aloud — this skill catches that on sight and rebuilds the piece in spoken form.

The philosophy in one line

A speech is **not an essay**. The reader of an essay can reread a sentence; the listener of a speech cannot. So the closing line matters more than the opening, the main thought must be repeated, and every sentence has to parse on first hearing.

Mode 1 — Draft from scratch

If the user gives you a topic + audience, walk through Kramon's drafting order **before writing a single body paragraph**:

1. **Write the closing line first.** One sentence. The line you want quoted. If you cannot write it, the user does not have a talk yet — ask them what they want the audience to remember tomorrow morning. 2. **Write the main thought.** One sentence. Different from the closing — the *argument*, not the *image*. Said early, repeated once mid-talk. 3. **Pick ONE story.** Real people, real names, one scene. If the user offers three stories, pick a lane — the other two are different talks. 4. **Pick the opener.** Three archetypes work: self-deprecating story, show-of-hands question, single concrete image. Pick the one that earns the laugh or the raised hand faster for *this* audience. Never *"Thank you for having me"* — throat-clearing. 5. **Outline the arc.** Opener → main thought → story → second beat or quote → restate main thought → closing line. 6. **Draft the body** in short paragraphs, the way the user actually speaks. Contractions. Dropped subjects where natural. Variable sentence length. 7. **Cut by one third.** Every talk runs long live. Pre-cut so it still fits. 8. **Read aloud at performance pace** with a timer. Mark stumbles — those are listener confusion points; rewrite.

Length target: **20 minutes max** for a keynote, **5–7 minutes** for a lightning talk, **2–3 minutes** for a toast or panel opener. Words per minute at speaking pace is ~120–140 for natural delivery — do the math both ways and don't trust either alone.

Mode 2 — Critique an existing draft

Mark the draft against the 12 rules. Quote the specific lines that fail and rewrite them inline. Flag any of these on sight:

  • **Opens with "Thank you for having me" / "I'm honored to be here" / "Let me start by saying"** — throat-clearing. Rewrite the opener as self-deprecating humor, audience question, or concrete image.
  • **No identifiable main thought in one sentence within the first two minutes** — the audience has no anchor. Ask the user what the main thought is; insert it as a standalone sentence early.
  • **More than one story** — pick a lane. Stories that don't serve the main thought are digressions that cost the room.
  • **Statistics without a story attached** — flip the order. Story first, then the stat that proves it. Stats alone are 12× less memorable than stories.
  • **Sentences over 25 words** — rewrite. The ear cannot hold a 40-word sentence.
  • **Nested clauses, jargon, undefined acronyms** — every one of these costs you the next sentence too. Plain word, plain sentence.
  • **Bullet-list slides** (if the user shows slides) — every bullet slide is a sentence the user did not finish writing. Force prose or kill the slide.
  • **Quotes longer than two sentences** — edit.
  • **Closes with "and with that, I'll take questions" / "thank you for your time"** — not an ending. Rewrite the close as a callback to the opener, a single quote, or a specific ask.
  • **Apologies for being underprepared, nervous, rushed** — never. Cut all of them.
  • **No repetition of the main thought** — say it once early, once mid, once at the close. Choose deliberately.
  • **The end is weaker than the opener** — flip the energy. The closing line is what they will quote.
  • **Three or more consecutive "what is" paragraphs** — the Sparkline is flat. Dense analytical stretches with no peak of wonder, humor, or rhetorical question lose the room. Mark each paragraph *is* or *could be*; insert a peak where three *is* run together. (See the Sparkline section of `points/speech-rules.md`.)

Mode 3 — Coaching the user through the read-aloud

If the user has the draft and is preparing to deliver, do not rewrite — coach the delivery. Ask:

1. Have you read it aloud at performance pace? (If no, do that first. Time it. Cut to fit.) 2. Have you memorized the first 30 seconds and the last 30 seconds word for word? (If no, do that next. The middle can improvise; the bookends cannot.) 3. Where do you stumble? (Stumbles are not the user's fault — they are listener confusion points. Rewrite those sentences plainer.) 4. What's the one slide / image that the audience will photograph? (If none, the talk needs a visual anchor.) 5. What's your fallback if the projector dies? (If none, the user is not

Read more
@kalyvask
Ships withwinning-writing

31 Claude skills for cold outreach, op-eds, pitches, press inquiries, bios, exec memos, performance reviews, spoken-delivery talks, fact-checking, and reply-rate tracking.

Stars on GitHub

Momentum

stars in total
17
Star history of kalyvask/winning-writing on GitHub, a chart by star-history.com

Other skills on winning-writing.

bluf-rewriter

bluf-rewriter

Reorganizes a memo, email, status update, or report so the bottom line is up front. Implements Kramon's BLUF (Bottom Line Up…

compression

compression

Cuts a draft to a target word count without losing substance. Two modes behind one skill. target-count hits a specific number…

fun-angle

fun-angle

Finds the unexpected, dry, or self-deprecating angle that makes an email memorable. Turns "professional and polite" into "this…

More skills from other plugins.

prompt-lookup

prompt-lookup

Activates when the user asks about AI prompts, needs prompt templates, wants to search for prompts, or mentions prompts.chat. Use…

@gh-f@gh-fInstall Skill
ppt-master

ppt-master

AI-driven presentation workflow for generating editable PPTX decks and slides, reconstructing page visuals, creating reusable…

@hugohe3@hugohe3Install Skill
morph-ppt

morph-ppt

Use this skill when the user wants a .pptx with smooth cross-slide animation — PowerPoint Morph transitions, Keynote-style…

@iofficeai@iofficeaiInstall Skill

Questions about speech-coach.

Which coding agents can use speech-coach?
speech-coach installs in Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and Windsurf with the skills CLI, using each agent's flag shown on this page. Flowy's auto-invocation, which fires a skill when your prompt matches, runs in Claude Code only.