agentleFS
TsukiNi22TsukiNi22/ai-utils/skills/ai/presentation/SKILL.md

presentation

Skill1 starsChanged 3 days agoMITView on GitHub

Make a presentation from a plan or an idea - a time (given, or estimated and confirmed) sets the number of slides; the format (PDF, one-file animated HTML, PPTX, several or all) and the style (several proposals for the subject, "White blueprint" by default, or copied from a PPTX / PDF / HTML the user gives) are asked through interactive-questions. One deck.json gives every format plus speaker notes. Use for "fais une présentation / des slides / un ppt / un pdf de présentation".

  • Installs packages

27 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use TubeScribe, deep-research-pro and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AlekseiULSkill

    TubeScribe

    YouTube video summarizer with speaker detection, formatted documents, and audio output. Works out of the box with macOS built-in TTS. Optional recommended tools (pandoc, ffmpeg, mlx-audio) enhance quality. Requires internet for YouTube access. No paid APIs or subscriptions. Use when user sends a YouTube URL or asks to summarize/transcribe a YouTube video.

    Also in 3 projects that use this

  • GPTtangSkill

    deep-research-pro

    Multi-source deep research agent. Searches the web, synthesizes findings, and delivers cited reports. No API keys required.

    Also in 3 projects that use this

  • vanzan01Skill

    tweet-writer

    Format instructions for writing tweets.

    Also in 3 projects that use this

  • vijpatel7Skill

    audit-website

    Audit websites for SEO, performance, security, technical, content, and 15 other issue cateories with 150+ rules using the squirrelscan CLI. Returns LLM-optimized reports with health scores, broken links, meta tag analysis, and actionable recommendations. Use to discover and asses website or webapp issues and health.

    Also in 3 projects that use this

  • ReScienceLabSkill

    reddit

    Search and retrieve content from Reddit. Get posts, comments, subreddit info, and user profiles via the public JSON API. Use when user mentions Reddit, a subreddit, or r/ links.

    Also in 3 projects that use this

  • sirakinbSkill

    last30days

    Research a topic from the last 30 days on Reddit + X + Web, become an expert, and write copy-paste-ready prompts for the user's target tool.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 24 edited versions we read keep these.

  • 3

    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:

    An idea / a subject: write the plan first (5-8 chapters, one line each) and show it before the slides.

    line 18
  2. Distinctive:

    Durée: - when the user gave a time, use it.

    line 26
  3. Very specific:

    Ask in the same question whether the file will be imported into Canva (or Google Slides / Keynote): then build with --canva (section 5b).

    line 33
  4. Very specific:

    Use preview to show a mini mock-up of each look (title line, colours, font) when it helps to choose.

    line 43
  5. Distinctive:

    No emoji, no decoration without meaning, nothing that moves unless it reveals the next point.

    line 115
  6. Very specific:

    Prefer the PPTX when the text must stay editable in Canva (it keeps one box per text and the table as a table).

    line 145

“Common” is measured: kept by at least 6 in 10 of the 24 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 (plan or idea -> slides in the formats asked)

SKILL_DIR = the folder of this file. One source, deck.json, built into an animated web page (one HTML file), a PDF (16:9, one page per slide), a PPTX and the speaker notes (<name>-notes.md), all in the same style. Everything written for the user's tools (scripts, JSON keys, file names) is English; the slides are in the language of the request (French here), code and comments in English.

1. What is given

  • A plan: the user's own outline (a file, a message): keep its order and its words, an idea / line of the plan is not forced to be one slide (one idea may take two or three slides, three small lines may share one).
  • An idea / a subject: write the plan first (5-8 chapters, one line each) and show it before the slides.
  • A model to follow (a PPTX / PDF / HTML / image of theirs): the style comes from it (section 3).
  • Ground it: read what the subject is about (the repository, the docs, the files named), and the earlier decks of the user when they exist (~/testing/ppt-tek, docs/, presentation/ folders): their vocabulary, their level, what they already explained. Never invent a number, a command or a source: what cannot be checked is marked "to check" in the notes.

2. Ask (ONE interactive-questions call, French, 3 questions)

  1. Durée - when the user gave a time, use it. Otherwise estimate it from the plan (about 1 content slide per minute, between 0.5 and 1: a talk of 20 minutes = 15 to 20 content slides, plus the title, the section dividers and the end) and ask whether it fits: options "<N> min, ~<S> slides (Recommended)", the same shorter, the same longer. The number of slides follows the time chosen; say it in the question.
  2. Format - options: PDF + HTML (Recommended) · PDF + HTML + PPTX (tout) · PPTX seulement · a custom mix through Other. The HTML is one self-contained file with build animations (the points appear one by one), the PDF is the same without motion (every step shown), the PPTX is editable (no animation) and carries the notes. Ask in the same question whether the file will be imported into Canva (or Google Slides / Keynote): then build with --canva (section 5b); the user can also say it in Other.
  3. Style - exactly 4 options, in this order:
    1. White blueprint (default) - recommended, always first, the label ends with "(default)".
    2. One other catalogue style that fits the subject (the "when" column below), different in mood.
    3. Sur mesure (sans template) - a look designed from scratch for this subject and this plan: its own palette, fonts, frame and background drawn from the subject's world, not taken from any catalogue theme (section 3, "Custom look"). Say in its description what it would look like (colours, mood) in a few words.
    4. Autre / custom - the user gives a file to copy ("copie le style de <fichier>", a PPTX / PDF / HTML / image) or describes the look in words; the free-text answer is read as such. Use preview to show a mini mock-up of each look (title line, colours, font) when it helps to choose.
  • Several decks asked in one go: one question call for all, not one per deck.
  • Unclear subject / audience / level: ask it in the same call (header "Public").

3. Styles

SKILL_DIR/templates/themes.json (and ~/.claude/presentation/themes/*.json, the user's own). Never hard-code a colour in a slide: the theme is the only place of the look.

ThemeLabelWhen
blueprintWhite blueprint (default)technical, code, architecture, repositories, courses: an architect's plan (off-white, fine blue grid, blue ink, title block)
paperClean paperany subject, reports, school defenses: neutral, readable
night-terminalNight terminalsystem, shell, security, low level: dark, monospace, green
warm-keynoteWarm keynotetalks, pitches, ideas: cream, orange, rounded, big words
slate-corporateSlate corporatebusiness case, company, formal defense: slate and deep blue
neon-darkNeon darkgames, creative, AI: deep violet, magenta and cyan

Custom look (Sur mesure) - the free, bold option: it is NOT limited to colours, a neon or a catalogue mood; the subject decides everything. Do not start from a catalogue row, start from an idea (a concept taken from the subject: a poster, a notebook, a terminal, a map, a comic page, a museum label, a blueprint of something else...) and push it: flashy colours or quiet ones, big type, hard shadows, full-colour section pages, giant numbers, tilted cards, gradients, patterns, drawn shapes (inline SVG in bgimg as url("data:image/svg+xml,...")), an unusual layout per kind. Two files in the deck folder, then --theme-file theme.json:

  • theme.json: every key of themes.json (label, dark, bg, paper, ink, ink2, tx, dim, code, codetx, warn, ok, font, head_font, mono, head_transform, head_spacing, head_weight, radius, bgimg, frame, inner_frame, cartouche) + "css_file": "theme.css" (or "css": "...").
  • theme.css: free CSS loaded after the layout (HTML and PDF): override anything (.k-title, .k-section, .col, .flow > li, .k-quote, #hud, @font-face, even transitions). Class names are in templates/slides.css (the stage, its scaling and the outline shell: templates/deck-base.css). templates/custom-example/ is a complete bold example ("Poster pop": cobalt, coral and yellow, hard shadows) to take the level of boldness from, not to copy. Rules that stay: text readable (contrast on every background, checked on the sheets), fonts that exist on the machine (fc-list), nothing leaving the slide (the selftest says it), no ::marker { content } (it crashes the PDF engine: use list-style-type). The PPTX carries the colours, fonts and background colour of the theme only (not the CSS), the --canva build ignores the CSS: say it. Offer to keep the result with the user's own themes (~/.claude/presentation/themes/, theme.css next to it).

Copy the style of a file the user gives (or "le style de la présentation sur le repo C++"):

python3 SKILL_DIR/scripts/extract_style.py <model.pptx|pdf|png|html|css> --name my-style [--save]

It prints a theme (colours and fonts read from the file). --save keeps it in ~/.claude/presentation/themes/ for the next decks. It is a first guess: build the example (templates/example.json) with --theme-file, LOOK at the sheets (section 5) and adjust the JSON (ink, bg, font, bgimg, frame...) until it looks like the model. Offer to set up a style once (--save) so that it is proposed in the next questions.

4. Write deck.json

SKILL_DIR/templates/example.json shows every kind. Keys: title, subtitle, lang, author, date, duration, theme, slides. A slide: kind, title, chapter (the bar of the web page), lead (one line under the title), notes (what is said aloud).

kindFieldsUse
titletitle, subtitle, metafirst slide
sectionnumber, title, leada chapter starts
bulletspoints: text or {text, sub}3-5 short points, shown one by one
columnscolumns: {head, points}compare 2-4 things
tablehead, rowsreal columns of data
codecode, lang, captiona command or a few lines (never a whole file)
stepssteps: {head, text}a path in 3-4 stages, revealed one by one
imagesrc (relative path), caption, alta screenshot, a photo, a QR code
svgsvg, captiona diagram drawn inline (currentColor = the text colour, <g style="color:var(--ink)"> for the accent, no fixed colour; the PDF resolves them from the theme); HTML / PDF only
quotetext, bythe sentence to remember
endtitle, subtitle, links, srclast slide: questions, links, QR code
Text accepts `code` and **bold**.

How a good deck is written (taken from the user's earlier decks and their remarks):

  • Less text, more visual: a diagram, a table, steps, a command instead of a paragraph; what is said aloud goes in notes, not on the slide. At most 5-6 points per slide, 60 words, one idea per slide.
  • Titles say the idea ("Un Makefile qui lance CMake"), not the topic ("Makefile").
  • Examples from the user's real projects and real commands, tested when they can be run.
  • A chapter order that answers the questions of someone new, in order; a recap or a quiz slide when the talk is a course.
  • The PDF is read without the speaker: it must be understandable alone, the HTML is for presenting.
  • A link worth keeping goes with a QR code image on the last slide (image / end with src).
  • No emoji, no decoration without meaning, nothing that moves unless it reveals the next point.

5. Build and check

python3 SKILL_DIR/scripts/build_deck.py deck.json -o <folder> --formats html,pdf,pptx,notes [--theme NAME | --theme-file F]
python3 SKILL_DIR/scripts/check_deck.py deck.json --dir <folder> [--sheets <tmpdir>]
  • check_deck.py: the pace (content slides per minute against duration), too many points / words, missing notes or images, the number of PDF pages; then contact sheets (PNG of every page of the PDF and of the PPTX rendered by LibreOffice) and the selftest of the web page (headless Firefox: text out of the slide, hidden by the bar, clipped, the keys). Open the PNGs (Read) and LOOK: a slide that overflows, a table too wide, a title on four lines, black on black. Fix the JSON (split the slide, shorten) and build again. Put the PNGs in /tmp, not in the folder delivered.
  • The web page keys: arrows / space (next step), Home / End, O (outline: chapter cards, the current slide marked, close with the cross or Esc; ?map in the address opens it), F (full screen), N (notes), swipe.
  • Tools: weasyprint (PDF), python-pptx (PPTX), firefox, pdftoppm, soffice (checks). A missing one is said with its install command; ask before installing (pip install --user python-pptx, sudo dnf install python3-python-pptx).

5b. Import into Canva (PDF or PPTX)

Canva rebuilds a PDF / PPTX as editable elements: it cuts a text where the font changes, drops a font it does not have, merges lines into one box, turns grouped shapes and gradients into one picture. So for a deck going to Canva:

python3 SKILL_DIR/scripts/build_deck.py deck.json -o <folder> --formats pdf,pptx --canva
python3 SKILL_DIR/scripts/check_deck.py deck.json --dir <folder> --canva
  • --canva keeps ONE family per role (body, headings, code) chosen in the Canva library (templates/canva.json, installed here so that it is embedded in the PDF), no fallback fonts, no letter spacing, no gradient grid background, no background / padding on inline code, no auto-fit in the PPTX. The look of the theme (colours, frames, layout) stays.
  • check_deck.py --canva reads the built files back: every string of every slide is found again as real text on its PDF page / in its PPTX shapes (a miss = text cut or turned into a picture), every font embedded and in the Canva list, no Type 3 font, no group or placeholder in the PPTX, no picture on a page that has none, no run without a font. An ERROR is fixed and rebuilt; a WARN is said to the user.
  • Prefer the PPTX when the text must stay editable in Canva (it keeps one box per text and the table as a table); the PDF when the look matters more. svg slides become pictures in Canva: use image or columns / steps for what must be edited.
  • What no script can check is how Canva itself imports it: tell the user to open it once and compare with the contact sheets; a font that Canva replaces anyway goes into templates/canva.json / out of it, then build again.

6. Deliver

  • Folder: the one the user names, otherwise presentation/<name>/ at the root of the current repository (or of the current folder): deck.json, <name>.html, <name>.pdf, <name>.pptx, <name>-notes.md, the images used.
  • Tell in French: the files, the time and number of slides, the style, what was not verified (a command not run, a figure to check), and what the PPTX does not carry (animations, inline drawings: give an image for those), and for a Canva import that the import itself was not tried.
  • A change asked afterwards ("plus court", "autre style"): edit deck.json, rebuild, no rewrite.

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.