| name | ideonomy-rich |
| description | Use when expanding an idea AND the output channel can render monospace cleanly โ terminals, READMEs, blog posts with monospace code blocks, fixed-width-font emails, ttyrec sessions, the Claude Code transcript itself. Same primitives as `ideonomy-plain`, but renders the artifact as performative ASCII art with Unicode box-drawing, figlet banners, density gradients, and visible ideonomy-machinery layers (tuple legend, dimensions surfaced, operator-named dividers, ideonomy trail). If the channel might mangle Unicode (Telegram, SMS, plain Slack DMs), use `ideonomy-plain` instead. |
Ideonomy-Rich
Sibling skill to ideonomy-plain. Same operators, organons, dimension-prompts, picker, cooldown. Different rendering policy.
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ ideonomy-plain โ lowest-common-denominator: survives SMS โ
โ ideonomy-rich โ highest-expression: feel ideas come alive โ
โ in a monospace terminal โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
If you're not sure which to pick, default to ideonomy-plain. Use this one when you know the medium can hold the art.
Core principle
Same as ideonomy-plain โ Patrick Gunkel's framework via Grace Kind. Ideas have properties varying along dimensions; expand by negating, substituting, combining, re-instantiating.
The difference is everything below: render the organon as performative ASCII art, not as plain bullets. Inspired by mahidalhan's ascii-art-explainer: composition teaches, motion shown, density is meaning, drama at the pivot.
How to invoke
Step 1 โ pick a method tuple.
PICK=$(find ~/.claude/plugins ~/.claude/skills -path '*/ideonomy-rich/bin/pick' -type f 2>/dev/null | head -1)
bash "$PICK"
(Same picker, same catalog as ideonomy-plain. Each skill carries its own copy so it drops cleanly into any skills/ folder on its own.)
Step 2 โ work through the tuple in two passes. Same internal pass as ideonomy-plain: dimension-prompts โ operators โ organon. The external pass is what changes.
Step 3 โ render the artifact in five visible layers.
A defining feature of ideonomy-rich: the brainstorming itself must be visible. The reader should be able to see which operator produced which section, what dimensions the brainstorming explored, and โ crucially โ what was not surfaced. Reading the artifact equals watching the ideonomy happen. If the reader can't tell the artifact came out of the operators ร organon ร dim-prompts, you buried the machinery.
This does not mean reverting to procedural headers (Phase A, Phase B). The middle path: name the operator AND the specific move it made, side by side, in content-named language.
Layer 1 โ title banner
Open with a figlet banner of the user's idea.
figlet -f slant "<idea>"
figlet -f doom "<idea>"
figlet -w 100 -f banner3 "<idea>"
toilet -f pagga --metal "<idea>"
curl -s "https://asciified.thelicato.io/api/v2/ascii?text=<idea>&font=Slant"
Recommended fonts by mood:
| Font | Mood |
|---|
slant | clean, modern, default |
doom | bold, declarative |
big | wide, readable |
banner3 | wide-display banner |
cyberlarge | tech / systems theme |
gothic | dramatic, weighty |
small | subtitles, secondary banners |
Frame the banner with hand-drawn โโโ if you want a heavy outer border, or leave it bare if the font has weight. Note: the boxes CLI's default designs render ASCII +--+ borders that clash with the Unicode aesthetic โ prefer hand-drawn โโโ framing or skip framing the banner.
Layer 2 โ tuple legend
Right after the banner, emit a compact block naming the tuple drawn for this artifact. The legend tells the reader "this is the brainstorming kit I used":
โญโ TUPLE โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โโฎ
โ โ OPERATORS substitution ยท negation โ
โ โ ORGANON dictionary โ
โ โ DIMENSIONS decomposability ยท naturalness ยท visibility โ
โฐโ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โ โโฏ
Use โญโโฎ rounded for the legend frame โ soft, indicating "metadata, not the artifact itself."
Layer 3 โ dimensions surfaced
Before the operator outputs, show what each picked dimension-prompt revealed about the user's idea. This is where the reader sees the axes the brainstorming explored. Each dim gets a small block: its axis (with โ / โ marking the values that matter), a one-line summary of what surfaced, and โ
on the pivot dim if one stands out:
โญโ DIMENSIONS โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ โ
โ decomposability monolithic โโโโโโโโโโโโโโโโโโโโโ decomposable โ
โ current is one big skill; alternatives split into โ
โ preflight + runtime + visible output โ
โ โ
โ naturalness instruction-as-discipline โโโโโโโโ artifact-as-discipline โ
โ discipline baked into the visible output, not the prompt โ
โ โ
โ visibility โ
invisible โโโโโโโโโโโโโโโโโโโโโโโ visible โ
โ 0% (current) โ footer-visible โ fully laid out โ
โ โ
pivot โ this is the user's stated complaint โ
โ โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
If the dim surfaced nothing useful for this idea, say so in one phrase rather than padding. Empty dims are a tell that the picker drew an unfit prompt; the reader benefits from seeing that.
Layer 4 โ artifact body, with operator-named dividers
The picked organon dictates the form. Each organon has a rendering recipe at ~/.claude/skills/ideonomy-rich/rendering/<organon>.md. Core recipes: chart, tree, list, atlas, scale, cycle, dictionary. Other organons fall back to the principles below.
The organon's name appears in the artifact's main header โ โโโ โ DICTIONARY โ โโโ followed by the body, never Phase C or Organon: dictionary.
When two or more operators were picked, divide their contributions with a divider that names BOTH the operator AND the specific move:
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ NEGATION โ negating "process is invisible" โ "process leaves a trail"
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
The divider has three visible parts:
- The operator's name (
NEGATION, SUBSTITUTION, COMBINATION, ABSTRACTION-LIFT, etc.)
- A one-line description of the specific move โ what was negated, substituted, combined
- The arrow
โ making the move visible
This is the difference that addresses the "where's the ideonomy?" complaint: the reader can see, at the divider, both the abstract operator and the concrete move it produced. โ OPPOSITES โ (content-only) hides the operator. Phase B โ Operator: Negation (procedure-only) hides the move. โ NEGATION โ "X" โ "Y" shows both.
Layer 5 โ ideonomy trail
Every artifact ends with a structured trail footer. The trail recaps the brainstorming machinery and โ most importantly โ names what was not surfaced. The "not surfaced" line is where the user finds adjacent directions of inquiry.
[ideonomy ยท 4 moves ยท 3 dims ยท 1 organon ยท "vigil v2 redesign"]
dim ยท pivot: visibility โ invisibleโvisible is the user's complaint
โ negation: "invisible by design" โ vigil-trail (Option A)
โ substitution: "self-grading" โ vigil-twin (Option B)
โ negation: "comprehensive" โ vigil-slim (Option C)
โ substitution: "single draft โ revise" โ vigil-tournament (Option D)
organon: dictionary โ 4 entries
not surfaced: "do nothing", "kill vigil entirely", "make vigil charge per-pass"
these are negations of meta-properties (existence, free-ness) the
picker's dim-prompts didn't reach. worth a follow-up tuple.
Trail rules:
- Always last in the artifact, in a single fenced code block, no trailing commentary
- One line per move; each line names operator + the specific input/output
- The
not surfaced: line is mandatory. Even one phrase. This is the line that turns the trail from a recap into a generator โ it tells the user where the next exploration could start.
- The trail makes the brainstorming falsifiable: if no trail appears, no real ideonomy ran (you wrote a thoughtful answer that happened to dress up nice).
Rendering principles
- Composition teaches, not labels. A flat box with text inside is wasted opportunity. Spatial arrangement, density, and motion should carry meaning. If your "diagram" is just labeled rectangles that could be bullet points, you've failed.
- Visual hierarchy through line weight. Mix
โโโ (single, default), โโโ (double, emphasis), and โญโโฎ (rounded, soft) within one artifact to mark primary / secondary / tertiary structure.
- Show motion.
โ โ โ โ โ โโโโถ ยทยทยท >>>>>> for flow, transformation, dependency. Static structures should still suggest direction where direction exists.
- Density is meaning.
โ โ โ โ shows intensity, fill, certainty, age, frequency. Pick a semantic axis and use the gradient consistently within one diagram.
- Decorative emphasis sparingly.
โ
โ โ โ โ โฒ โผ for markers, status, importance. One symbol per role within a piece โ โ always means empty/coinage opportunity, โ always means canonical instance, etc.
- Drama at the pivot. If the expansion has a key insight or a single empty cell that's the point, let composition draw the eye there โ center it, frame it with
ยซ ยป or โฆ โง, isolate it with whitespace. Make the reader land on it.
Character palette
Box (single): โ โ โ โ โ โ โ โค โฌ โด โผ
Box (double): โ โ โ โ โ โ โ โฃ โฆ โฉ โฌ
Box (rounded): โญ โ โฎ โ โฐ โฏ
Density: โ โ โ โ โ โ โ โ
Geometric: โ โ โ โ โ โ โ โก โฒ โณ โผ โฝ โ
โ โฆ โง
Arrows: โ โ โ โ โ โ โ โ โถ โต โโโโถ โโโโ โป โบ
Diagonals: โฑ โฒ
Brackets: โฆ โง โจ โฉ ยซ ยป โ โ โ โ
Constraints
- Width โค 100 chars per line. Most terminals default to 80; 100 leaves a comfortable margin while letting wide art breathe. Anything wider wraps and turns ugly.
- Wrap large diagrams in fenced code blocks (
```). Even a monospace-friendly channel may proportional-font your prose; the fence guarantees alignment.
- One large pyfiglet banner per artifact (the title). Use
small/mini for sub-banners if needed; don't make every section a 12-line banner.
- No ANSI color in saved-text contexts.
toilet --gay looks great in a live terminal, looks like \e[31m garbage in a markdown file. Color belongs to live tty only.
- Box-drawing styles do not mix within a single diagram. Pick one of
โโโ / โโโ / โญโโฎ and stick with it; mixing +--+ ASCII with โโโ Unicode is the cardinal sin.
Red flags
Brainstorming-machinery invisibility (the big one):
- No tuple legend after the banner โ reader can't see which kit was used.
- No
DIMENSIONS block โ the axes the brainstorming explored are hidden.
- Dividers say
โ OPPOSITES โ (content-only, no operator named) โ the machinery is invisible; this is the failure mode that prompted Layer 4 to require operator+move.
- No ideonomy trail at the end โ no falsifiability; the artifact is indistinguishable from a thoughtful answer that didn't use ideonomy at all.
- Trail's
not surfaced: line missing or padded with filler โ the most generative line is the most likely to be skipped; treat its absence as a vigil-style "trail-without-vigil" tell.
Procedure leaking out (don't over-correct):
- A section titled
Phase A/B/C, Operator: <name>, Organon: <name> โ procedure leaking. The fix is not to drop the operator name from the divider; it's to pair it with the specific move (โ NEGATION โ "X" โ "Y").
Aesthetic / formatting:
- Plain bullets and prose, no organon visually rendered โ you skipped the rendering layer.
- Figlet banner so wide it wraps in 100-char terminal โ use
small or shorten.
- Decorative symbols (
โ
โโ) sprayed without consistent semantic role โ noise.
- Visual hierarchy doesn't track conceptual hierarchy โ most important = most visually emphatic.
- Mixed
+--+ and โโโ in one diagram โ pick one style.
- The artifact is just labels-in-boxes that could have been a bullet list โ composition isn't teaching anything.
See also
ideonomy-plain โ sibling skill, plain-text-portable. Same primitives, different rendering policy. Install it alongside this one if you want both available.rendering/ (within this skill) โ per-organon ASCII recipes.- Hermes
ascii-art skill โ pyfiglet, cowsay, boxes, image-to-ascii. Install if missing.
- mahidalhan/claude-hacks
ascii-art-explainer โ performative ASCII philosophy this skill inherits.
- mahidalhan/claude-hacks
ascii-explainer โ diagnose-then-render approach for diagrams.
