make any deck look designed with one free skill | LAB 0025 field notes


Lenny's AI Builders - LAB 0025

the free skill that tied Claude Slides in my blind test, how to install it, and the scores.

if you want a deck that looks designed without opening a design tool, there's a free GitHub skill for Claude Code called frontend-slides. you paste your notes, pick a style, and get one web page deck.

i put it up against Claude Slides and ChatGPT Slides on the same client pitch. a blind judge scored all six decks. frontend-slides averaged 6 out of 10, the same as Claude Slides. ChatGPT Slides averaged 4.

why it matters

most AI decks look like AI decks. this one gives you a styled, single-file web page you can present from a browser or export as a PDF, and it's free and MIT licensed.

it also runs inside Claude Code, so it can read your notes, files or a call transcript straight from your project.

what you get

  • a deck that looks designed. it shows you three styles and you pick one.
  • one file. the whole deck is a single web page, easy to share or host.
  • a PDF if you need one. ask it to export the deck.

the real example: my result

on 20 October 2026 i gave the same one page of client notes (a 10-slide pitch for a strata company) to Claude Slides, ChatGPT Slides and Claude Code with frontend-slides, two runs each. my agent ran the decks for me.

all six decks were shuffled into letters A to F with the names hidden. a judge from a different AI (Gemini 3.1 Pro) scored them twice, once forwards and once backwards, before i looked.

  • 1st: Claude Slides (run 2), 6.5
  • joint 2nd: frontend-slides, both runs, 6.0 each
  • 4th: Claude Slides (run 1), 5.5
  • last: ChatGPT Slides, both runs, 4.0 each

per tool: Claude Slides 6.0, frontend-slides 6.0, ChatGPT Slides 4.0. the two judging passes disagreed on the single winner: one picked Claude Slides, the other picked a frontend-slides deck. ChatGPT lost, so i cancelled my ChatGPT plan live in the video.

try it

  1. open Claude Code.
  2. add the marketplace: /plugin marketplace add https://github.com/zarazhangrui/frontend-slides
  3. install the skill: /plugin install frontend-slides@frontend-slides
  4. run it: type /frontend-slides and pick it from the list, then paste the prompt below with your notes.
  5. pick one of the three styles it shows you.
  6. open the deck in your browser. ask for a PDF if you need a file.

copy this prompt

paste this after /frontend-slides, then paste your notes underneath:

Build a 10-slide deck from the notes below. It's a speaking deck for a client pitch, so keep the text on each slide short.
Show me your three style options first, then build the one I pick.
Save the final deck as a single file called deck.html.
[paste your notes or a call transcript here]

the install and how to use it starts at 10:09 in the episode.

you'll know it worked when

you get one deck.html file that opens in your browser, has the number of slides you asked for, and says what your notes say.

a bad result: slides that invent facts that weren't in your notes, or walls of text. tell it what's wrong and ask it to redo those slides.

what to do next

  • got a pitch this week: paste your call notes in and present straight from the browser.
  • already pay for Claude: try Claude Slides on the same notes too. in my test the two were neck and neck.
  • need to send a file: export the PDF, and keep the web version for presenting.

limits and dates

  • small test. two runs per tool on one brief, 20 October 2026. say "on my test", not "the best".
  • the judge saw still frames only, so animations didn't count.
  • the PDF export screenshots each slide, so you lose the animations in the PDF.
  • needs Claude Code. the skill is free, but you need a Claude plan to run Claude Code.
  • ChatGPT Slides ran on a Pro account, because Slides wasn't on Plus when i tested.

the rest of this LAB

  • Claude Slides is in every Claude chat, even on the free plan. ChatGPT Slides just got live collaboration on Pro.
  • Meta's Muse Glimmer goes open. an open model you can run on your own computer.
  • Anthropic's Public Benefit Fund, and Elon Musk's reply.
  • Claude's free OSS Scanner for open source projects.
  • Grok Bot on the iPhone Action Button.
  • Tibo's day 14 update, and The Information's report on Anthropic's next model.

sources

next step

watch LAB 0025 on YouTube - the skill starts at 10:09, the blind reveal at 12:03.

made a deck with it? tell me under the video which style you picked.

want the next one? get the next field notes. optional - these notes stay free to read.

Lenny’s AI Builders

i’m Lennox. i sift through AI Twitter and share one thing i tried with ChatGPT. for people making products, content and useful systems. five emails a week, with a weekly option.

Read more from Lenny’s AI Builders

Lenny's AI Builders - LAB 0045 one Claude Code setting, 36 to 38% fewer tokens in every pair of my test, the catch on cost, and a prompt that gives you 10 versions. if you build with Claude Code, one new setting can make it use fewer tokens. i tested this myself. every pair of runs used 36 to 38% fewer tokens. but there is a catch. my first run cost 83% more at api prices. the other two pairs came out 15% and 17% cheaper. why it matters the setting turns on Code Mode. Claude writes code that...

Lenny's AI Builders - LAB 0042 the one Grok Bot upgrade that matters, and the paragraph i add to claude.md for long Fable 5.2 jobs. Grok Bot 1.0 isn't new, but the launch post pulled 46 million views like it was. so don't worry, you're not missing out. the one upgrade that's actually decent got something like 40 times fewer views. and it let my bot walk into my Mac. why it matters Grok Bot has been out since August. xAI's own page calls 1.0 a major update to its AI teammates. the headline...

Lenny's AI Builders - LAB 0041 the conductor paragraph, the four steps to run it, and the honest numbers from my 40-hour run. i ran my viral Fable 5.2 conductor for 40 hours on my real app. one conductor, 73 worker threads, 25 pull requests merged. and my weekly Fable limit was gone in 13 hours. so here's what it actually built, and the one paragraph that runs yours. why it matters this is the new meta. you use Fable 5.2 as the conductor and Opus 5.5 as the worker threads. Fable 5.2 is twice...