dososo GitHub avatar

blcaptain-ppt-skill

dososo

An AI-agent Skill that turns ideas, articles, or datasets into single-file HTML presentations with machine-enforced design taste and anti-fabrication honesty across 7 visual personas.

Stars

31

7-day growth

No data

Forks

3

Open issues

0

License

NOASSERTION

Last updated

2026-06-27

AI repository intelligence
FR-AI / ANALYSIS

Why it is worth attention

It replaces luck-driven AI slides with a rigorously enforced quality floor (WCAG contrast, spacing, 32-dim audit) and prohibits fabricated data, emoji, and stock photos, anchoring each persona to a recognized design system rather than recolored templates.

Who it is for

  • AI agent users (Codex, Claude Code, Cursor, Gemini CLI)
  • Technical professionals preparing reports or keynotes
  • Data analysts needing honest, data-centric presentations
  • Product teams creating premium or editorial-style decks

Use cases

  • Convert a technical article or research paper into a presentation
  • Generate a data-driven industry analysis deck with charts
  • Create a product launch keynote with a luxury or minimalist persona
  • Produce an editorial-style long-form analysis for internal or external use

Strengths

  • Machine-enforced consistency: every deck passes WCAG AAA contrast, spacing gates, and a 32-dimension audit before shipping
  • Zero external dependencies: uses only Node built-ins, no npm packages
  • Anti-fabrication honesty: unverifiable numbers are marked 'illustrative' and enforced by machine, no fake logos or screenshots
  • 7 distinct personas each anchored to a proven design system (e.g., Tufte, Constructivism, De Stijl), not recolored templates

Considerations

  • Not suitable for large data tables, training materials, or legal/medical/financial compliance documents
  • Does not support multi-author editing or PPTX export natively (planned)
  • Requires a Skill-capable agent and Node ≥18; commercial use of the software itself needs a paid license

README quick start

blcaptain-ppt-skill

Turn an idea / article / dataset into single-file HTML presentations across 7 visual personas — where both taste and honesty are machine-enforced

English · 中文

Install — tell your agent (Codex / Claude Code / Cursor / Gemini CLI…): "Install this Skill: github.com/dososo/blcaptain-ppt-skill"


The idea — Engineered Conviction

Most AI slides announce themselves at a glance: big color blocks, emoji, purple gradients, centered stacking, made-up numbers, random stock photos. The problem isn't that the model isn't trying — it's that "looking good" is treated as mysticism, left to luck, different every time.

But real presentation design — a technical report, a launch keynote, an industry analysis — has a method: how much the type scale should differ, how much whitespace to leave, which colors you must never use, how a dark slide stays readable on a projector, which numbers you may not invent. Recognized design systems have studied this for decades.

blcaptain turns that craft into constraints a machine can enforce. Every deck is a precisely-engineered object that carries one judgment you can't dodgeprecision × conviction. The AI doesn't free-style a layout; it fills a validated one. So the output is stably good, and honest — every time, not by luck.

The moat — what we refuse to do

"Good design is as little design as possible." — Dieter Rams. The moat isn't a feature list; it's the discipline of what we won't ship.

  • Never fabricate data / logos / screenshots / sources — unverifiable numbers are labeled illustrative, and it's machine-checked.
  • No emoji, no random stock photos — the two loudest tells of AI slop.
  • One signal color, ~5% — no multi-color clutter fighting for attention.
  • No decorative geometry that carries no meaning — form serves content, never the reverse.
  • No frivolous font weights, no all-rounded floating cards, no purple/blue/orange/pink gradients.
  • Never crowd out whitespace to cram more in.
  • No unanchored, self-invented styles — every persona anchors to a recognized design system, with sources.
  • No copying a single school wholesale as a finished product.

What you get

7 visual personasEach anchored to a recognized design system (not recolored

Description

AI 原生·单文件 HTML 演示 Skill:7 套锚定公认设计体系的视觉人格,好看(WCAG/间距/32 维审计)与诚实(反伪造)都由机器强制,零依赖。 | AI-native single-file HTML presentation skill — 7 design-system-anchored visual personas with machine-enforced taste and an anti-fabrication discipline. Zero dependencies.

Related repositories

Similar projects matched by category, topics, and programming language.

ddosi
Featured
ddosi GitHub avatar

PacketLens

PacketLens is a pure front-end, offline pcap analysis tool that runs entirely in the browser, supporting deep protocol decoding, HTTPS decryption, and million-packet instant loading without any backend.

HTML
17
ViffyGwaanl
Featured
ViffyGwaanl GitHub avatar

kimi-k3-learn

An interactive learning system that turns the 47-page Kimi K3 technical report into a single offline HTML file with running algorithms, 3D visualizations, spaced repetition quizzes, and a smart highlighting QA tool.

AI & Machine Learning
15
iamtechartist
Featured
iamtechartist GitHub avatar

human-cell-visualizer

An interactive 3D visualization of three human cell types using Three.js, WebGL particles, and custom GLSL shaders, allowing users to rotate, zoom, morph, and explore annotated structures.

Design & Creative
13