ljg-present

Original：🇨🇳 Chinese

Translated

Takahashi Method Presentation Generator. Distill the ideas of a text into beats, one keyword per page, with text filling the entire screen. Go full-screen in the browser to start presenting. Users may trigger it with phrases like: 'Takahashi', 'present this', 'present', 'make a speech', 'create a Takahashi PPT', 'present this', 'show it', 'forge a presentation', 'make slides', 'turn this into a Takahashi presentation'. Outputs a single HTML file to ~/Downloads/.

16installs

Sourcelijigang/ljg-skills

Added on2026-04-22

NPX Install

npx skill4agent add lijigang/ljg-skills ljg-present

SKILL.md Content (Chinese)

View Translation Comparison →

ljg-present: Takahashi Method Presentation Forging

Cut text into beats, reduce PPT to a metronome, and return the stage to the speaker.

What This Is NOT

Not PPT creation — Takahashi Method is anti-PPT
Not summarization — summarization is abbreviation, Takahashi Method is beats
Not an outline — outlines emphasize fluency, Takahashi Method emphasizes jumps
Not a decorative show — no images, no icons at all

What This Is

Origin of Takahashi Method: In 2001, Japanese Ruby programmer Masayuki Takahashi wrote a few pages of large-font slides in HTML when no professional presentation tools were available. It unexpectedly became popular and evolved into a genre.

Three iron rules — violating any means it's not Takahashi Method:

One idea unit per page — not a paragraph per page, but the "word that should ring out at this moment"
Text fills the entire screen — visually, only these words remain on the screen; no white space for "breathing", the words themselves are the breath
No images, tables, or decorations — the speaker is the protagonist, PPT is the metronome. Metronomes don't need frills

Core Philosophy

PPT is a metronome, not an information carrier.

Traditional PPT: Information-dense, audience reads the screen, speaker reads off the PPT. Takahashi Method: Information-sparse, audience listens to the speaker, PPT keeps the beat.

This means each distilled page is not an "abbreviation", but the "word that should flash in the audience's mind at this moment during the speaker's narration".

Distillation Rules

Distillation ≠ Summarization

Summarization (Wrong)	Distillation (Right)
Compress 500 words into 100 words	Compress 500 words into "the core point the author is really driving home"
Retain information	Retain markers
Continuous transition	Parallel jumps
Save reading time	Let the speaker think of this word exactly during the presentation

Six-Step Distillation Method

Step 1: Read through to find the core point — What is the core point the author is really driving home? 1, at most 3 — The core point is "the point they keep returning to", not the title or topic sentence

Step 2: Arrange the rhythm — Around the core point, arrange 8-24 "idea markers" — Each marker is the "click" moment in the audience's mind — Too few pages: Can't support a full presentation; Too many pages: Lose the beat feeling

Step 3: Compress text to ≤ 12 characters — Use single characters if possible — Use phrases if possible — Use nouns instead of verbs — Use nouns instead of sentences

Step 4: Strip modifiers and connectors — Remove adjectives (unless the adjective itself is the marker) — Remove "therefore/so/but/however" — these are the speaker's responsibility — Remove "we/you/everyone" — Takahashi Method has no "we" — Remove exclamation marks, question marks — visuals don't shout; let the speaker do that

Step 5: Check the jump feeling — Adjacent pages should have a "click-click" dislocated feeling, not a "smooth-smooth" continuous feeling — If reading two merged pages makes sense, it means the cut isn't sharp enough — A single page must stand on its own to be qualified

Step 6: Keep the "concluding page" — The last page is a conclusion, not a summary — A single character, a symbol, or a rhetorical phrase — The speaker pauses on that page, and it resonates in the audience's mind

Distillation Example

Original Text (160 words):

We always think we're making choices, but most of the time we're on default. Did you choose this company because it's really better than others? Or because HR contacted you first? Did you choose your daily wake-up time because you thought it through? Or because your phone alarm is set there? Default is a choice. But it's a choice made for you by others. The first step to living differently from others is realizing you're being defaulted.

Takahashi Method Distillation (12 pages):

1.  Choice
2.  Most of the time
3.  On default
4.  Default ≠ Choice
5.  Who chose for you
6.  HR contacted first
7.  Alarm rang first
8.  Chosen by others
9.  First step
10. Realize
11. Being defaulted
12. Wake up

Note: The last page "Wake up" is a single character. This is the conclusion.

Takahashi Method Visual Specifications

Hard Constraints

Font size large enough — Use
```
clamp()
```
+
```
vh
```
to automatically fill the viewport based on word count
No images, icons, illustrations, tables, borders
No transition animations — Page turns are hard cuts
No headers or footers — Only page numbers (small gray text, bottom left corner)

Soft Constraints (Color Choices for This Skill)

Background color: Washi paper cream
```
#F7F3E9
```
Main text: Ink black
```
#1A1A1A
```
Accent color: Vermilion red
```
#C03C28
```
(Only used for concluding pages or words the author wants to emphasize)

Font stack:

"PingFang SC", "Heiti SC", "STHeiti", -apple-system, sans-serif

This doesn't violate "no decorations" — color itself creates a sense of markers, not decoration.

Calling Process

Step 1: Obtain Content

URL → Fetch via
```
WebFetch
```
Pasted text → Use directly
File path → Fetch via
```
Read
```

Step 2: Distill (Follow Six-Step Method)

Distill the text into a JSON array following the six-step method. Pages with

emphasis: true

are rendered in vermilion red (usually only 1-2 pages in the whole presentation, mostly the conclusion).

json

[
  {"text": "Choice", "emphasis": false},
  {"text": "Most of the time", "emphasis": false},
  {"text": "On default", "emphasis": false},
  ...
  {"text": "Wake up", "emphasis": true}
]

Step 3: Render

Read

assets/takahashi_template.html

, replace

{{SLIDES_JSON}}

with the JSON string above, and replace

{{TITLE}}

with the presentation topic (used for browser tab title, ≤ 20 characters).

Step 4: Write File

Extract the topic from the content as

{name}

(Chinese, no punctuation, ≤ 20 characters):

~/Downloads/{name}-Takahashi.html

Step 5: Deliver

Report the absolute path, inform the user:

Double-click to open in browser
Use
```
→
```
```
←
```
```
Space
```
to turn pages
Press
```
f
```
for full screen
Use
```
Home
```
/
```
End
```
to jump to first/last page

Taste Guidelines

Single-character pages are treasures — At least 1-2 single-character pages in the whole presentation; these are where the beat is loudest
No obsession with parallelism — Pages don't need aligned word counts; the beat comes from variation
Concluding page stands alone — The last page is the soul; vermilion red is only for it
Numbers can be pages — Words like "ten years", "zero", "3%" are naturally Takahashi Method words
Punctuation is restrained — Basically no punctuation. Split a sentence into two pages, and it naturally breaks
English can be interspersed — If the original word is English, keep it as is, don't translate (e.g., "default", "YES")

Chinese Default

Default output is Chinese. Unless the original text is English and the user requests to keep it in English.

Forbidden Zones (Violating These Means It's Not Takahashi Method)

No multi-line pages — One line per page, at most two lines
No bullet points — Takahashi Method has no lists
No tables — Tables are data, data is not a beat
No charts / images / emoji icons — Pure text
No gradients / shadows / transition animations — Hard cuts
No meta-language like "Key points of this page" — Don't explain, only present