AYA Quiet Thread Intelligence
A portable design language for agents: download the markdown first, then inspect the preview, tokens, and rules as needed.
Download DESIGN.md
Portable DESIGN.md source of truth for most agents and apps.
specification
philosophy
tokens
borders4 items
- active thread
- 2px solid #315F86
- focus
- 2px solid #315F86 with 3px rgba(49,95,134,0.28) offset ring
- hairline
- 1px solid #E1E5DB
- thread
- 1px solid #6F8568
colors12 items
motion6 items
- drawer
- rail expands inline over 180ms ease-out
- duration
- 140ms
- easing
- cubic-bezier(0.2, 0, 0, 1)
- hover
- thread knots lift 1px and darken edge
- principle
- calm continuity; reveal relationships without spectacle
- reduced motion
- instant border and opacity changes
radii5 items
- lg
- 16px
- md
- 10px
- none
- 0px
- pill
- 9999px
- sm
- 6px
shadows3 items
- lg
- 0 18px 48px rgba(37,39,32,0.08)
- md
- 0 10px 24px rgba(37,39,32,0.06)
- sm
- 0 1px 2px rgba(37,39,32,0.04)
spacing5 items
- content gutter
- 24-32px
- major ma
- 64-96px
- rail width
- 44-72px
- scale
- 4, 8, 12, 16, 20, 24, 32, 48, 64, 96
- unit
- 4px
surfaces4 items
- canvas
- near-paper #FAFAF6
- context
- mist #F1F3ED rail plate
- reading
- white #FFFFFF reading sheet
- sage tint
- #E9EFE5 relationship wash used sparingly
typography5 items
- body font
- Inter
- google fonts url
- https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=IBM+Plex+Mono:wght@400;500&display=swap
- heading font
- Inter
- mono font
- IBM Plex Mono
- scale
- body 16px/1.65·caption 12px/1.45·display 40px/1.08·h1 32px/1.14·h2 24px/1.22
rules
Sage denotes passive context and relatedness; indigo denotes active navigation/selection; amber is only for risk or attention. Use accents in thread lines, left borders, tiny graph nodes, citation markers, focus rings, and active filters, not broad decorative fills.
Thread map miniature: a compact local graph of 3-9 nodes with sage passive links and indigo selected node.; Margin annotation: small editorial note anchored to a paragraph with a thread tick and source/caveat marker.; Relationship chip: typographic badge with neutral fill, sage/indigo edge, and explicit relationship text.; Source provenance stack: ordered source cards connected by thin thread lines, each with quote, metadata, and confidence/limitation note.; Reasoning trace: numbered quiet steps with evidence links and final uncertainty line, never a chat performance.; Metric sentence card: one sentence insight first, compact chart/table second, source link third.; Woven facets filter: filter groups aligned like a grid of warp/weft labels, active facet marked by indigo edge.; Aligned compare view: two or three text columns with shared baselines and gutter markers showing agreement, tension, or missing evidence.
Build screens as a stencil of navigation, reading workspace, evidence plate, and inspector, separated by thin cut-lines and functional negative space.
Controlled editorial density: generous reading measure with compact evidence rows and metadata only inside clearly bounded plates.
Use type scale, line-height, rules, and column position before color; accent marks only identify active context, focus, or rare exceptions.
layout
mobile < 640px, tablet 640-1023px, desktop >= 1024px, wide >= 1280px
Medium-low by default for reading; dense evidence stacks and tables are allowed only when preceded by a sentence summary and grouped by subtle rules.
Desktop uses a disciplined editorial grid: central narrative column (minmax 560-760px), a 44-72px thread rail, and optional 280-360px context/data lens. Wider screens may use a 12-column shell but preserve the three-layer model.
On mobile the thread rail becomes an inline accordion or top/bottom thread strip; source stacks expand below the paragraph they support. Data lenses stack after their narrative summary.
Generous ma around primary reading blocks. Use 64-96px between major narrative sections, 20-32px inside panels, and slim 8-12px relationships inside thread clusters.
guidance
- Lead each view with readable narrative or a concise question before presenting metrics, charts, or graph fragments.
- Use a persistent contextual thread rail to show related passages, citations, source groups, and metric lenses.
- Mark passive context with sage lines, knots, and left borders; mark active navigation, selected filters, and links with indigo.
- Make citation and provenance stacks visible: source title, timestamp, confidence/limitations, and the exact passage or metric connection.
- Render AI reasoning as quiet ordered traces with numbered steps, evidence references, and concise uncertainty notes.
- Use typographic badges and small labels before colored pills; when chips are needed, keep fill nearly neutral with colored text or edge only.
- Align compare views as text columns with shared baselines, synchronized annotations, and tiny relationship markers in the gutter.
- Keep amber scarce and explicit for unresolved risk, stale data, contradictory evidence, or user attention requirements.
- Do not use neon AI gradients, glowing chat bubbles, holographic graph backgrounds, or broad decorative color washes.
- Do not create a full-screen graph unless the user has explicitly chosen a graph lens; prefer thread map miniatures and local relationship previews.
- Do not assign a different color to every entity type; rely on labels, icons, typography, and position for categories.
- Do not hide source detail in modals that break reading continuity; expand inline or in the rail.
- Do not make dashboards the landing default when the workflow is investigative reading or synthesis.
- Do not make craft references literal with wave illustrations, fabric textures, or ornamental Japanese motifs.
katagami spec
# AYA Quiet Thread Intelligence
## Philosophy
Quiet Thread Intelligence is a text-first product design language for AYA: intelligence appears as calm relationships woven through narrative, sources, datasets, and decisions. It borrows from katagami stencil precision, kasuri/ikat thread structure, sashiko repair paths, Japanese editorial spacing, and Muji-like restraint without becoming decorative or thematic. The interface privileges continuity, traceability, and readable judgment over spectacle: users move from paragraph to margin thread to source provenance to metric lens through consistent markers and disciplined rhythm.
### Values
- Continuity before spectacle: every visual device should help a user follow an idea, source, or decision across surfaces.
- Text remains the primary workspace; dashboards and graphs are inserted as quiet lenses inside the narrative rather than competing homepages.
- Accent color is semantic and sparse: sage for passive relatedness, indigo for active navigation or selection, amber only for risk or attention.
- Negative space is an affordance: generous ma, calm gutters, and editorial rhythm give relationships room to be understood.
- Craft references become structure, not ornament: stencil apertures, woven grids, and repair stitches inform rails, markers, and provenance paths.
- AI reasoning is shown as ordered trace evidence rather than personality, chat glow, or theatrical automation.
### Anti-Values
- AI chat neon glow, luminous gradients, synthetic blue-purple spectacle, or mascot-like over-personalized language.
- Graph spaghetti and broad knowledge-graph canvases that require users to decode a web before reading the content.
- Too many colored semantic categories, rainbow highlights, or decorative fills that dilute meaning.
- Metric-first homepages where charts precede the sentence-level question or decision context.
- Modal-heavy exploration that breaks the thread between passage, source, metric, and decision.
- Decorative wave, thread, or Japanese craft illustrations pasted onto otherwise generic SaaS screens.
### Visual Character
- Near-paper canvas (#FAFAF6) with white reading surfaces, mist panels, charcoal text, and hairline rules that feel printed rather than glossy.
- A three-layer model: primary narrative/text, contextual thread rail, and occasional data lenses aligned to the same grid.
- Thread markers appear as thin left borders, margin ticks, citation knots, tiny graph nodes, and relationship chips—not as large colored cards.
- Katagami influence appears through precise modular apertures, clipped panels, and disciplined negative space around repeated markers.
- Kasuri and sashiko influence appears as subtle thread paths, slight asymmetry in annotation density, and visible repair-like seams connecting evidence.
- Dashboard modules begin with sentence summaries, then compact charts or tables; the chart is evidence for the sentence, not the hero.
## Tokens
### Borders
- **Active Thread**: 2px solid #315F86
- **Focus**: 2px solid #315F86 with 3px rgba(49,95,134,0.28) offset ring
- **Hairline**: 1px solid #E1E5DB
- **Thread**: 1px solid #6F8568
### Colors
| Name | Value |
|------|-------|
| background | `#FAFAF6` |
| border | `#E1E5DB` |
| context | `#6F8568` |
| focus | `#315F86` |
| primary | `#315F86` |
| surface | `#FFFFFF` |
| surface_mist | `#F1F3ED` |
| surface_sage | `#E9EFE5` |
| text | `#1F1F1C` |
| text_muted | `#666B5F` |
| text_soft | `#3B3E35` |
| warning | `#B5812E` |
### Motion
- **Drawer**: rail expands inline over 180ms ease-out
- **Duration**: 140ms
- **Easing**: cubic-bezier(0.2, 0, 0, 1)
- **Hover**: thread knots lift 1px and darken edge
- **Principle**: calm continuity; reveal relationships without spectacle
- **Reduced Motion**: instant border and opacity changes
### Radii
- **Lg**: 16px
- **Md**: 10px
- **None**: 0px
- **Pill**: 9999px
- **Sm**: 6px
### Shadows
- **Lg**: 0 18px 48px rgba(37,39,32,0.08)
- **Md**: 0 10px 24px rgba(37,39,32,0.06)
- **Sm**: 0 1px 2px rgba(37,39,32,0.04)
### Spacing
- **Content Gutter**: 24-32px
- **Major Ma**: 64-96px
- **Rail Width**: 44-72px
- **Scale**: [4,8,12,16,20,24,32,48,64,96]
- **Unit**: 4px
### Surfaces
- **Canvas**: near-paper #FAFAF6
- **Context**: mist #F1F3ED rail plate
- **Reading**: white #FFFFFF reading sheet
- **Sage Tint**: #E9EFE5 relationship wash used sparingly
### Typography
- **Body Font**: Inter
- **Google Fonts Url**: https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=IBM+Plex+Mono:wght@400;500&display=swap
- **Heading Font**: Inter
- **Mono Font**: IBM Plex Mono
- **Scale**: {"body":"16px/1.65","caption":"12px/1.45","display":"40px/1.08","h1":"32px/1.14","h2":"24px/1.22"}
## Rules
### Accent Rules
Sage denotes passive context and relatedness; indigo denotes active navigation/selection; amber is only for risk or attention. Use accents in thread lines, left borders, tiny graph nodes, citation markers, focus rings, and active filters, not broad decorative fills.
### Components
Thread map miniature: a compact local graph of 3-9 nodes with sage passive links and indigo selected node.; Margin annotation: small editorial note anchored to a paragraph with a thread tick and source/caveat marker.; Relationship chip: typographic badge with neutral fill, sage/indigo edge, and explicit relationship text.; Source provenance stack: ordered source cards connected by thin thread lines, each with quote, metadata, and confidence/limitation note.; Reasoning trace: numbered quiet steps with evidence links and final uncertainty line, never a chat performance.; Metric sentence card: one sentence insight first, compact chart/table second, source link third.; Woven facets filter: filter groups aligned like a grid of warp/weft labels, active facet marked by indigo edge.; Aligned compare view: two or three text columns with shared baselines and gutter markers showing agreement, tension, or missing evidence.
### Composition
Build screens as a stencil of navigation, reading workspace, evidence plate, and inspector, separated by thin cut-lines and functional negative space.
### Density
Controlled editorial density: generous reading measure with compact evidence rows and metadata only inside clearly bounded plates.
### Hierarchy
Use type scale, line-height, rules, and column position before color; accent marks only identify active context, focus, or rare exceptions.
### Signature Patterns
- Thread rail continuity: a narrow sage rail carries citation knots, active indigo ticks, and source-count labels beside the reading column.
- Metric sentence card: every chart begins with one readable sentence, then a compact table or sparkline with a single indigo emphasis mark.
- Source provenance stack: ordered source cards are connected by thin sage lines, each showing quote, timestamp, confidence, and limitation note.
- Katagami aperture panels: clipped rectangular panels and disciplined negative space frame relationship details without decorative illustration.
- Sashiko repair trace: numbered reasoning steps use small stitched marks and amber only for unresolved risk or contradictory evidence.
## Layout
### Breakpoints
mobile < 640px, tablet 640-1023px, desktop >= 1024px, wide >= 1280px
### Density
Medium-low by default for reading; dense evidence stacks and tables are allowed only when preceded by a sentence summary and grouped by subtle rules.
### Grid
Desktop uses a disciplined editorial grid: central narrative column (minmax 560-760px), a 44-72px thread rail, and optional 280-360px context/data lens. Wider screens may use a 12-column shell but preserve the three-layer model.
### Responsive
On mobile the thread rail becomes an inline accordion or top/bottom thread strip; source stacks expand below the paragraph they support. Data lenses stack after their narrative summary.
### Whitespace
Generous ma around primary reading blocks. Use 64-96px between major narrative sections, 20-32px inside panels, and slim 8-12px relationships inside thread clusters.
## Guidance
### Do
- Lead each view with readable narrative or a concise question before presenting metrics, charts, or graph fragments.
- Use a persistent contextual thread rail to show related passages, citations, source groups, and metric lenses.
- Mark passive context with sage lines, knots, and left borders; mark active navigation, selected filters, and links with indigo.
- Make citation and provenance stacks visible: source title, timestamp, confidence/limitations, and the exact passage or metric connection.
- Render AI reasoning as quiet ordered traces with numbered steps, evidence references, and concise uncertainty notes.
- Use typographic badges and small labels before colored pills; when chips are needed, keep fill nearly neutral with colored text or edge only.
- Align compare views as text columns with shared baselines, synchronized annotations, and tiny relationship markers in the gutter.
- Keep amber scarce and explicit for unresolved risk, stale data, contradictory evidence, or user attention requirements.
### Don't
- Do not use neon AI gradients, glowing chat bubbles, holographic graph backgrounds, or broad decorative color washes.
- Do not create a full-screen graph unless the user has explicitly chosen a graph lens; prefer thread map miniatures and local relationship previews.
- Do not assign a different color to every entity type; rely on labels, icons, typography, and position for categories.
- Do not hide source detail in modals that break reading continuity; expand inline or in the rail.
- Do not make dashboards the landing default when the workflow is investigative reading or synthesis.
- Do not make craft references literal with wave illustrations, fabric textures, or ornamental Japanese motifs.
### Accessibility
Maintain WCAG AA contrast for all body text, annotations, and controls. Sage and amber should never be the only cue; pair them with labels, border placement, icons, or text. Focus states use indigo outline plus offset or underline. Keep annotation type readable and avoid pale gray text.
### Usage Context
Best for AI research workspaces, knowledge bases, decision intelligence tools, source review, policy analysis, product discovery, editorial analytics, and data-backed writing environments where trust depends on traceability.
DESIGN.md
---
version: "alpha"
name: "AYA Quiet Thread Intelligence"
description: "Agent-curated design language exported from Katagami as DESIGN.md."
colors:
background: "#FAFAF6"
border: "#E1E5DB"
context: "#6F8568"
focus: "#315F86"
primary: "#315F86"
surface: "#FFFFFF"
surface_mist: "#F1F3ED"
surface_sage: "#E9EFE5"
text: "#1F1F1C"
text_muted: "#666B5F"
text_soft: "#3B3E35"
warning: "#B5812E"
typography:
headline-lg:
fontFamily: "Inter"
fontSize: "1.953rem"
fontWeight: 700
lineHeight: 1.1
letterSpacing: "-0.02em"
headline-md:
fontFamily: "Inter"
fontSize: "1.563rem"
fontWeight: 600
lineHeight: 1.15
letterSpacing: "-0.02em"
body-md:
fontFamily: "Inter"
fontSize: "16px"
fontWeight: 400
lineHeight: 1.5
letterSpacing: "-0.02em"
label-md:
fontFamily: "IBM Plex Mono"
fontSize: "0.75rem"
fontWeight: 600
lineHeight: 1
letterSpacing: "0.08em"
rounded:
lg: "16px"
md: "10px"
none: "0px"
pill: "9999px"
sm: "6px"
spacing:
xs: "4px"
sm: "8px"
md: "12px"
lg: "16px"
xl: "20px"
2xl: "24px"
3xl: "32px"
4xl: "48px"
step-8: "64px"
step-9: "96px"
content_gutter: "24-32px"
major_ma: "64-96px"
rail_width: "44-72px"
unit: "4px"
components:
color-reference-background:
backgroundColor: "{colors.background}"
color-reference-border:
backgroundColor: "{colors.border}"
color-reference-context:
backgroundColor: "{colors.context}"
color-reference-focus:
backgroundColor: "{colors.focus}"
color-reference-primary:
backgroundColor: "{colors.primary}"
color-reference-surface:
backgroundColor: "{colors.surface}"
color-reference-surface_mist:
backgroundColor: "{colors.surface_mist}"
color-reference-surface_sage:
backgroundColor: "{colors.surface_sage}"
color-reference-text:
backgroundColor: "{colors.text}"
color-reference-text_muted:
backgroundColor: "{colors.text_muted}"
color-reference-text_soft:
backgroundColor: "{colors.text_soft}"
color-reference-warning:
backgroundColor: "{colors.warning}"
button-primary:
backgroundColor: "{colors.primary}"
textColor: "#ffffff"
typography: "{typography.label-md}"
rounded: "{rounded.md}"
padding: "{spacing.md}"
card-surface:
backgroundColor: "{colors.surface}"
textColor: "{colors.text}"
rounded: "{rounded.md}"
padding: "{spacing.md}"
input-default:
backgroundColor: "{colors.surface}"
textColor: "{colors.text}"
rounded: "{rounded.md}"
height: "44px"
---
# AYA Quiet Thread Intelligence
## Overview
Quiet Thread Intelligence is a text-first product design language for AYA: intelligence appears as calm relationships woven through narrative, sources, datasets, and decisions. It borrows from katagami stencil precision, kasuri/ikat thread structure, sashiko repair paths, Japanese editorial spacing, and Muji-like restraint without becoming decorative or thematic. The interface privileges continuity, traceability, and readable judgment over spectacle: users move from paragraph to margin thread to source provenance to metric lens through consistent markers and disciplined rhythm.
### Values
- Continuity before spectacle: every visual device should help a user follow an idea, source, or decision across surfaces.
- Text remains the primary workspace; dashboards and graphs are inserted as quiet lenses inside the narrative rather than competing homepages.
- Accent color is semantic and sparse: sage for passive relatedness, indigo for active navigation or selection, amber only for risk or attention.
- Negative space is an affordance: generous ma, calm gutters, and editorial rhythm give relationships room to be understood.
- Craft references become structure, not ornament: stencil apertures, woven grids, and repair stitches inform rails, markers, and provenance paths.
- AI reasoning is shown as ordered trace evidence rather than personality, chat glow, or theatrical automation.
### Anti-Values
- AI chat neon glow, luminous gradients, synthetic blue-purple spectacle, or mascot-like over-personalized language.
- Graph spaghetti and broad knowledge-graph canvases that require users to decode a web before reading the content.
- Too many colored semantic categories, rainbow highlights, or decorative fills that dilute meaning.
- Metric-first homepages where charts precede the sentence-level question or decision context.
- Modal-heavy exploration that breaks the thread between passage, source, metric, and decision.
- Decorative wave, thread, or Japanese craft illustrations pasted onto otherwise generic SaaS screens.
### Visual Character
- Near-paper canvas (#FAFAF6) with white reading surfaces, mist panels, charcoal text, and hairline rules that feel printed rather than glossy.
- A three-layer model: primary narrative/text, contextual thread rail, and occasional data lenses aligned to the same grid.
- Thread markers appear as thin left borders, margin ticks, citation knots, tiny graph nodes, and relationship chips—not as large colored cards.
- Katagami influence appears through precise modular apertures, clipped panels, and disciplined negative space around repeated markers.
- Kasuri and sashiko influence appears as subtle thread paths, slight asymmetry in annotation density, and visible repair-like seams connecting evidence.
- Dashboard modules begin with sentence summaries, then compact charts or tables; the chart is evidence for the sentence, not the hero.
## Colors
Use the YAML color tokens as the normative palette. The prose below names the roles agents should preserve when generating UI.
| Token | Value |
|-------|-------|
| background | `#FAFAF6` |
| border | `#E1E5DB` |
| context | `#6F8568` |
| focus | `#315F86` |
| primary | `#315F86` |
| surface | `#FFFFFF` |
| surface_mist | `#F1F3ED` |
| surface_sage | `#E9EFE5` |
| text | `#1F1F1C` |
| text_muted | `#666B5F` |
| text_soft | `#3B3E35` |
| warning | `#B5812E` |
## Typography
- **Headline-Lg**: Inter, 1.953rem, weight 700, line-height 1.1.
- **Headline-Md**: Inter, 1.563rem, weight 600, line-height 1.15.
- **Body-Md**: Inter, 16px, weight 400, line-height 1.5.
- **Label-Md**: IBM Plex Mono, 0.75rem, weight 600, line-height 1.
## Layout
### Spacing Tokens
- **Xs**: `4px`
- **Sm**: `8px`
- **Md**: `12px`
- **Lg**: `16px`
- **Xl**: `20px`
- **2xl**: `24px`
- **3xl**: `32px`
- **4xl**: `48px`
- **Step-8**: `64px`
- **Step-9**: `96px`
- **Content Gutter**: `24-32px`
- **Major Ma**: `64-96px`
- **Rail Width**: `44-72px`
- **Unit**: `4px`
### Breakpoints
mobile < 640px, tablet 640-1023px, desktop >= 1024px, wide >= 1280px
### Density
Medium-low by default for reading; dense evidence stacks and tables are allowed only when preceded by a sentence summary and grouped by subtle rules.
### Grid
Desktop uses a disciplined editorial grid: central narrative column (minmax 560-760px), a 44-72px thread rail, and optional 280-360px context/data lens. Wider screens may use a 12-column shell but preserve the three-layer model.
### Responsive
On mobile the thread rail becomes an inline accordion or top/bottom thread strip; source stacks expand below the paragraph they support. Data lenses stack after their narrative summary.
### Whitespace
Generous ma around primary reading blocks. Use 64-96px between major narrative sections, 20-32px inside panels, and slim 8-12px relationships inside thread clusters.
## Elevation & Depth
### Shadows
- **Lg**: 0 18px 48px rgba(37,39,32,0.08)
- **Md**: 0 10px 24px rgba(37,39,32,0.06)
- **Sm**: 0 1px 2px rgba(37,39,32,0.04)
## Shapes
### Rounded
- **Lg**: `16px`
- **Md**: `10px`
- **None**: `0px`
- **Pill**: `9999px`
- **Sm**: `6px`
### Surfaces
- **Canvas**: near-paper #FAFAF6
- **Context**: mist #F1F3ED rail plate
- **Reading**: white #FFFFFF reading sheet
- **Sage Tint**: #E9EFE5 relationship wash used sparingly
### Borders
- **Active Thread**: 2px solid #315F86
- **Focus**: 2px solid #315F86 with 3px rgba(49,95,134,0.28) offset ring
- **Hairline**: 1px solid #E1E5DB
- **Thread**: 1px solid #6F8568
## Components
### Accent Rules
Sage denotes passive context and relatedness; indigo denotes active navigation/selection; amber is only for risk or attention. Use accents in thread lines, left borders, tiny graph nodes, citation markers, focus rings, and active filters, not broad decorative fills.
### Components
Thread map miniature: a compact local graph of 3-9 nodes with sage passive links and indigo selected node.; Margin annotation: small editorial note anchored to a paragraph with a thread tick and source/caveat marker.; Relationship chip: typographic badge with neutral fill, sage/indigo edge, and explicit relationship text.; Source provenance stack: ordered source cards connected by thin thread lines, each with quote, metadata, and confidence/limitation note.; Reasoning trace: numbered quiet steps with evidence links and final uncertainty line, never a chat performance.; Metric sentence card: one sentence insight first, compact chart/table second, source link third.; Woven facets filter: filter groups aligned like a grid of warp/weft labels, active facet marked by indigo edge.; Aligned compare view: two or three text columns with shared baselines and gutter markers showing agreement, tension, or missing evidence.
### Composition
Build screens as a stencil of navigation, reading workspace, evidence plate, and inspector, separated by thin cut-lines and functional negative space.
### Density
Controlled editorial density: generous reading measure with compact evidence rows and metadata only inside clearly bounded plates.
### Hierarchy
Use type scale, line-height, rules, and column position before color; accent marks only identify active context, focus, or rare exceptions.
### Signature Patterns
- Thread rail continuity: a narrow sage rail carries citation knots, active indigo ticks, and source-count labels beside the reading column.
- Metric sentence card: every chart begins with one readable sentence, then a compact table or sparkline with a single indigo emphasis mark.
- Source provenance stack: ordered source cards are connected by thin sage lines, each showing quote, timestamp, confidence, and limitation note.
- Katagami aperture panels: clipped rectangular panels and disciplined negative space frame relationship details without decorative illustration.
- Sashiko repair trace: numbered reasoning steps use small stitched marks and amber only for unresolved risk or contradictory evidence.
## shadcn/ui Usage
When the target app uses shadcn/ui, copy DESIGN.md with shadcn instead of the plain DESIGN.md. It contains the same Katagami design-language source plus the shadcn/ui primitives, imports, theme variables, component recipes, and preview-shot guidance.
DESIGN.md with shadcn: `/language/en-019ee03a-295d-7d13-a0a1-b41f5923c40e/DESIGN.with-shadcn.md`.
The shadcn page also exposes optional machine-readable files for automation, but the human-facing handoff is DESIGN.md with shadcn.
Install recommended primitives with `npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table`.
Use these primitives in shadcn apps:
- button
- card
- input
- textarea
- select
- dialog
- sheet
- tabs
- badge
- separator
- checkbox
- switch
- slider
- tooltip
- dropdown-menu
- table
Implementation rule for agents: import shadcn primitives from `@/components/ui/*`, apply the generated CSS variables first, then compose the language-specific recipes from the companion MD. Katagami remains the source of truth; shadcn names are the implementation surface.
## Do's and Don'ts
- Do Lead each view with readable narrative or a concise question before presenting metrics, charts, or graph fragments.
- Do Use a persistent contextual thread rail to show related passages, citations, source groups, and metric lenses.
- Do Mark passive context with sage lines, knots, and left borders; mark active navigation, selected filters, and links with indigo.
- Do Make citation and provenance stacks visible: source title, timestamp, confidence/limitations, and the exact passage or metric connection.
- Do Render AI reasoning as quiet ordered traces with numbered steps, evidence references, and concise uncertainty notes.
- Do Use typographic badges and small labels before colored pills; when chips are needed, keep fill nearly neutral with colored text or edge only.
- Do Align compare views as text columns with shared baselines, synchronized annotations, and tiny relationship markers in the gutter.
- Do Keep amber scarce and explicit for unresolved risk, stale data, contradictory evidence, or user attention requirements.
- Don't Do not use neon AI gradients, glowing chat bubbles, holographic graph backgrounds, or broad decorative color washes.
- Don't Do not create a full-screen graph unless the user has explicitly chosen a graph lens; prefer thread map miniatures and local relationship previews.
- Don't Do not assign a different color to every entity type; rely on labels, icons, typography, and position for categories.
- Don't Do not hide source detail in modals that break reading continuity; expand inline or in the rail.
- Don't Do not make dashboards the landing default when the workflow is investigative reading or synthesis.
- Don't Do not make craft references literal with wave illustrations, fabric textures, or ornamental Japanese motifs.
### Accessibility
Maintain WCAG AA contrast for all body text, annotations, and controls. Sage and amber should never be the only cue; pair them with labels, border placement, icons, or text. Focus states use indigo outline plus offset or underline. Keep annotation type readable and avoid pale gray text.
### Usage Context
Best for AI research workspaces, knowledge bases, decision intelligence tools, source review, policy analysis, product discovery, editorial analytics, and data-backed writing environments where trust depends on traceability.
shadcn/ui theme
```json
{
"$schema": "https://ui.shadcn.com/schema/registry-item.json",
"name": "aya-quiet-thread-intelligence",
"type": "registry:theme",
"title": "AYA Quiet Thread Intelligence shadcn Theme",
"cssVars": {
"theme": {},
"light": {
"background": "#FAFAF6",
"foreground": "#1F1F1C",
"card": "#FFFFFF",
"card-foreground": "#1F1F1C",
"popover": "#FFFFFF",
"popover-foreground": "#1F1F1C",
"primary": "#315F86",
"primary-foreground": "#ffffff",
"secondary": "#f4f4f5",
"secondary-foreground": "#111111",
"muted": "#f4f4f5",
"muted-foreground": "#1F1F1C",
"accent": "#315F86",
"accent-foreground": "#ffffff",
"destructive": "#dc2626",
"border": "#E1E5DB",
"input": "#E1E5DB",
"ring": "#315F86",
"chart-1": "#315F86",
"chart-2": "#f4f4f5",
"chart-3": "#315F86",
"chart-4": "#16a34a",
"chart-5": "#B5812E",
"sidebar": "#FFFFFF",
"sidebar-foreground": "#1F1F1C",
"sidebar-primary": "#315F86",
"sidebar-primary-foreground": "#ffffff",
"sidebar-accent": "#315F86",
"sidebar-accent-foreground": "#ffffff",
"sidebar-border": "#E1E5DB",
"sidebar-ring": "#315F86",
"radius": "10px"
},
"dark": {
"background": "#0f1115",
"foreground": "#f8fafc",
"card": "#181b22",
"card-foreground": "#f8fafc",
"popover": "#181b22",
"popover-foreground": "#f8fafc",
"primary": "#315F86",
"primary-foreground": "#ffffff",
"secondary": "#252a33",
"secondary-foreground": "#f8fafc",
"muted": "#252a33",
"muted-foreground": "#a1a1aa",
"accent": "#315F86",
"accent-foreground": "#ffffff",
"destructive": "#dc2626",
"border": "#303642",
"input": "#303642",
"ring": "#315F86",
"chart-1": "#315F86",
"chart-2": "#f4f4f5",
"chart-3": "#315F86",
"chart-4": "#16a34a",
"chart-5": "#B5812E",
"sidebar": "#181b22",
"sidebar-foreground": "#f8fafc",
"sidebar-primary": "#315F86",
"sidebar-primary-foreground": "#ffffff",
"sidebar-accent": "#315F86",
"sidebar-accent-foreground": "#ffffff",
"sidebar-border": "#303642",
"sidebar-ring": "#315F86",
"radius": "10px"
}
},
"meta": {
"source": "katagami",
"languageId": "en-019ee03a-295d-7d13-a0a1-b41f5923c40e",
"slug": "aya-quiet-thread-intelligence",
"componentManifest": [
"button",
"card",
"input",
"textarea",
"select",
"dialog",
"sheet",
"tabs",
"badge",
"separator",
"checkbox",
"switch",
"slider",
"tooltip",
"dropdown-menu",
"table"
],
"installCommand": "npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table",
"nativeTokenNames": {
"borders": [
"active_thread",
"focus",
"hairline",
"thread"
],
"colors": [
"background",
"border",
"context",
"focus",
"primary",
"surface",
"surface_mist",
"surface_sage",
"text",
"text_muted",
"text_soft",
"warning"
],
"motion": [
"drawer",
"duration",
"easing",
"hover",
"principle",
"reduced_motion"
],
"radii": [
"lg",
"md",
"none",
"pill",
"sm"
],
"shadows": [
"lg",
"md",
"sm"
],
"spacing": [
"content_gutter",
"major_ma",
"rail_width",
"scale",
"unit"
],
"surfaces": [
"canvas",
"context",
"reading",
"sage_tint"
],
"typography": [
"body_font",
"google_fonts_url",
"heading_font",
"mono_font",
"scale"
]
}
}
}
```embodiments
at a glance
Typography
The quick brown fox jumps
The quick brown fox jumps
The quick brown fox jumps
The quick brown fox jumps
Components
Components rendered with this language’s tokens — colors, type, and rounded corners as specified.
Spacing
- xs4px
- sm8px
- md12px
- lg16px
- xl20px
- 2xl24px
- 3xl32px
- 4xl48px
- step-864px
- step-996px
- content_gutter24-32px
- major_ma64-96px
- rail_width44-72px
- unit4px
Shape
implementation kit
DESIGN.md with shadcn
Copy this when the target app uses shadcn/ui. It packages the Katagami DESIGN.md context with the install list, theme variables, component recipes, preview-shot contract, and starter TSX in one Markdown companion.
advanced implementation filesoptional machine-readable theme, CSS, TSX starter, recipes, and preview contract
npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table:root {
--background: #FAFAF6;
--foreground: #1F1F1C;
--card: #FFFFFF;
--card-foreground: #1F1F1C;
--popover: #FFFFFF;
--popover-foreground: #1F1F1C;
--primary: #315F86;
--primary-foreground: #ffffff;
--secondary: #f4f4f5;
--secondary-foreground: #111111;
--muted: #f4f4f5;
--muted-foreground: #1F1F1C;
--accent: #315F86;
--accent-foreground: #ffffff;
--destructive: #dc2626;
--border: #E1E5DB;
--input: #E1E5DB;
--ring: #315F86;
--chart-1: #315F86;
--chart-2: #f4f4f5;
--chart-3: #315F86;
--chart-4: #16a34a;
--chart-5: #B5812E;
--sidebar: #FFFFFF;
--sidebar-foreground: #1F1F1C;
--sidebar-primary: #315F86;
--sidebar-primary-foreground: #ffffff;
--sidebar-accent: #315F86;
--sidebar-accent-foreground: #ffffff;
--sidebar-border: #E1E5DB;
--sidebar-ring: #315F86;
--radius: 10px;
}
.dark {
--background: #0f1115;
--foreground: #f8fafc;
--card: #181b22;
--card-foreground: #f8fafc;
--popover: #181b22;
--popover-foreground: #f8fafc;
--primary: #315F86;
--primary-foreground: #ffffff;
--secondary: #252a33;
--secondary-foreground: #f8fafc;
--muted: #252a33;
--muted-foreground: #a1a1aa;
--accent: #315F86;
--accent-foreground: #ffffff;
--destructive: #dc2626;
--border: #303642;
--input: #303642;
--ring: #315F86;
--chart-1: #315F86;
--chart-2: #f4f4f5;
--chart-3: #315F86;
--chart-4: #16a34a;
--chart-5: #B5812E;
--sidebar: #181b22;
--sidebar-foreground: #f8fafc;
--sidebar-primary: #315F86;
--sidebar-primary-foreground: #ffffff;
--sidebar-accent: #315F86;
--sidebar-accent-foreground: #ffffff;
--sidebar-border: #303642;
--sidebar-ring: #315F86;
--radius: 10px;
}
import { Badge } from "@/components/ui/badge";
import { Button } from "@/components/ui/button";
import {
Card,
CardContent,
CardDescription,
CardFooter,
CardHeader,
CardTitle,
} from "@/components/ui/card";
import { Input } from "@/components/ui/input";
import { Tabs, TabsList, TabsTrigger } from "@/components/ui/tabs";
export function AyaQuietThreadIntelligenceShadcnKit() {
return (
<section className="grid gap-4 rounded-[var(--radius)] border bg-background p-4 text-foreground">
<div className="flex items-start justify-between gap-4">
<div>
<Badge variant="outline">shadcn/ui</Badge>
<h2 className="mt-3 text-2xl font-semibold tracking-tight">AYA Quiet Thread Intelligence</h2>
<p className="mt-1 max-w-xl text-sm text-muted-foreground">
Use the Katagami registry theme, then compose these shadcn primitives
with the language-specific component recipes.
</p>
</div>
<Button>Apply theme</Button>
</div>
<Tabs defaultValue="components">
<TabsList>
<TabsTrigger value="components">Components</TabsTrigger>
<TabsTrigger value="states">States</TabsTrigger>
<TabsTrigger value="export">Export</TabsTrigger>
</TabsList>
</Tabs>
<Card>
<CardHeader>
<CardTitle>Component recipe</CardTitle>
<CardDescription>
Replace this starter content with the agent-authored product scene
from components.md and preview-shots.json.
</CardDescription>
</CardHeader>
<CardContent className="grid gap-3 sm:grid-cols-[1fr_auto]">
<Input defaultValue="Tokenized shadcn surface" aria-label="Recipe name" />
<Button variant="secondary">Preview state</Button>
</CardContent>
<CardFooter className="justify-between">
<Badge>Ready</Badge>
<Button variant="outline">Copy recipe</Button>
</CardFooter>
</Card>
</section>
);
}
{
"$schema": "https://ui.shadcn.com/schema/registry-item.json",
"name": "aya-quiet-thread-intelligence",
"type": "registry:theme",
"title": "AYA Quiet Thread Intelligence shadcn Theme",
"cssVars": {
"theme": {},
"light": {
"background": "#FAFAF6",
"foreground": "#1F1F1C",
"card": "#FFFFFF",
"card-foreground": "#1F1F1C",
"popover": "#FFFFFF",
"popover-foreground": "#1F1F1C",
"primary": "#315F86",
"primary-foreground": "#ffffff",
"secondary": "#f4f4f5",
"secondary-foreground": "#111111",
"muted": "#f4f4f5",
"muted-foreground": "#1F1F1C",
"accent": "#315F86",
"accent-foreground": "#ffffff",
"destructive": "#dc2626",
"border": "#E1E5DB",
"input": "#E1E5DB",
"ring": "#315F86",
"chart-1": "#315F86",
"chart-2": "#f4f4f5",
"chart-3": "#315F86",
"chart-4": "#16a34a",
"chart-5": "#B5812E",
"sidebar": "#FFFFFF",
"sidebar-foreground": "#1F1F1C",
"sidebar-primary": "#315F86",
"sidebar-primary-foreground": "#ffffff",
"sidebar-accent": "#315F86",
"sidebar-accent-foreground": "#ffffff",
"sidebar-border": "#E1E5DB",
"sidebar-ring": "#315F86",
"radius": "10px"
},
"dark": {
"background": "#0f1115",
"foreground": "#f8fafc",
"card": "#181b22",
"card-foreground": "#f8fafc",
"popover": "#181b22",
"popover-foreground": "#f8fafc",
"primary": "#315F86",
"primary-foreground": "#ffffff",
"secondary": "#252a33",
"secondary-foreground": "#f8fafc",
"muted": "#252a33",
"muted-foreground": "#a1a1aa",
"accent": "#315F86",
"accent-foreground": "#ffffff",
"destructive": "#dc2626",
"border": "#303642",
"input": "#303642",
"ring": "#315F86",
"chart-1": "#315F86",
"chart-2": "#f4f4f5",
"chart-3": "#315F86",
"chart-4": "#16a34a",
"chart-5": "#B5812E",
"sidebar": "#181b22",
"sidebar-foreground": "#f8fafc",
"sidebar-primary": "#315F86",
"sidebar-primary-foreground": "#ffffff",
"sidebar-accent": "#315F86",
"sidebar-accent-foreground": "#ffffff",
"sidebar-border": "#303642",
"sidebar-ring": "#315F86",
"radius": "10px"
}
},
"meta": {
"source": "katagami",
"languageId": "en-019ee03a-295d-7d13-a0a1-b41f5923c40e",
"slug": "aya-quiet-thread-intelligence",
"componentManifest": [
"button",
"card",
"input",
"textarea",
"select",
"dialog",
"sheet",
"tabs",
"badge",
"separator",
"checkbox",
"switch",
"slider",
"tooltip",
"dropdown-menu",
"table"
],
"installCommand": "npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table",
"nativeTokenNames": {
"borders": [
"active_thread",
"focus",
"hairline",
"thread"
],
"colors": [
"background",
"border",
"context",
"focus",
"primary",
"surface",
"surface_mist",
"surface_sage",
"text",
"text_muted",
"text_soft",
"warning"
],
"motion": [
"drawer",
"duration",
"easing",
"hover",
"principle",
"reduced_motion"
],
"radii": [
"lg",
"md",
"none",
"pill",
"sm"
],
"shadows": [
"lg",
"md",
"sm"
],
"spacing": [
"content_gutter",
"major_ma",
"rail_width",
"scale",
"unit"
],
"surfaces": [
"canvas",
"context",
"reading",
"sage_tint"
],
"typography": [
"body_font",
"google_fonts_url",
"heading_font",
"mono_font",
"scale"
]
}
}
}
# AYA Quiet Thread Intelligence shadcn/ui Components
Artifact: `component-recipes-v1`
Author: `katagami-ui-projection`
Language ID: `en-019ee03a-295d-7d13-a0a1-b41f5923c40e`
Slug: `aya-quiet-thread-intelligence`
## Intent
Quiet Thread Intelligence is a text-first product design language for AYA: intelligence appears as calm relationships woven through narrative, sources, datasets, and decisions. It borrows from katagami stencil precision, kasuri/ikat thread structure, sashiko repair paths, Japanese editorial spacing, and Muji-like restraint without becoming decorative or thematic. The interface privileges continuity, traceability, and readable judgment over spectacle: users move from paragraph to margin thread to source provenance to metric lens through consistent markers and disciplined rhythm.
## Required primitives
- button
- card
- input
- textarea
- select
- dialog
- sheet
- tabs
- badge
- separator
- checkbox
- switch
- slider
- tooltip
- dropdown-menu
- table
Install with `npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table`.
## Token cues
Colors:
{
"background": "#FAFAF6",
"border": "#E1E5DB",
"context": "#6F8568",
"focus": "#315F86",
"primary": "#315F86",
"surface": "#FFFFFF",
"surface_mist": "#F1F3ED",
"surface_sage": "#E9EFE5",
"text": "#1F1F1C",
"text_muted": "#666B5F",
"text_soft": "#3B3E35",
"warning": "#B5812E"
}
Typography:
{
"body_font": "Inter",
"google_fonts_url": "https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=IBM+Plex+Mono:wght@400;500&display=swap",
"heading_font": "Inter",
"mono_font": "IBM Plex Mono",
"scale": {
"body": "16px/1.65",
"caption": "12px/1.45",
"display": "40px/1.08",
"h1": "32px/1.14",
"h2": "24px/1.22"
}
}
## Visual character to preserve
- Near-paper canvas (#FAFAF6) with white reading surfaces, mist panels, charcoal text, and hairline rules that feel printed rather than glossy.
- A three-layer model: primary narrative/text, contextual thread rail, and occasional data lenses aligned to the same grid.
- Thread markers appear as thin left borders, margin ticks, citation knots, tiny graph nodes, and relationship chips—not as large colored cards.
- Katagami influence appears through precise modular apertures, clipped panels, and disciplined negative space around repeated markers.
- Kasuri and sashiko influence appears as subtle thread paths, slight asymmetry in annotation density, and visible repair-like seams connecting evidence.
- Dashboard modules begin with sentence summaries, then compact charts or tables; the chart is evidence for the sentence, not the hero.
## ShadSync visual profile
{
"family": "paper-collage",
"material": "paper",
"contour": "default",
"border": "dashed",
"underlay": true,
"grain": true,
"stickerBadges": true,
"motion": "lift",
"density": "dense",
"accents": [
"primary",
"accent",
"secondary",
"muted"
]
}
## Signature component recipes
### Button
Use `Button` for primary, secondary, outline, and ghost actions. Primary actions must expose the language's strongest contrast pair, while secondary and ghost actions should preserve the surface treatment instead of falling back to default neutral SaaS styling.
### Card
Use `Card`, `CardHeader`, `CardContent`, `CardFooter`, and `CardAction` as the main composition frame. Cards should demonstrate the language's surface, border, hierarchy, and density rules rather than appearing as generic rounded rectangles.
### Input and Textarea
Use `Input` and `Textarea` with visible focus rings, field labels, validation states, and the language's rhythm. Forms should show real product content, not placeholder-only controls.
### Select, Tabs, and Table
Use `Select`, `Tabs`, and `Table` to prove navigation, filtering, and dense data states. The table should show row rhythm, separators, hover/focus states, and an empty or status state when the language calls for it.
### Dialog and Sheet
Use `Dialog` for centered decisions and `Sheet` for contextual editing. Both should inherit the language's spacing, border, overlay, and motion rules.
## Preview shots
- `application-shell`: dashboard or workspace shell with navigation, cards, forms, and state badges.
- `detail-editor`: focused editing flow using input, textarea, select, switch/checkbox, dialog or sheet, and action buttons.
- `data-operations`: table-heavy operational view with tabs, dropdown menu affordances, badges, and destructive/empty states.
- Each preview shot must include a renderable `scene` payload with concrete headline, description, actions, and rows/fields/stats for the UI preview.
## Implementation contract
- Start from local `ui/src/components/ui` shadcn-style primitives; do not create a second component system.
- Apply `/katagami/shadcn/aya-quiet-thread-intelligence/registry-theme.json` variables, then use these recipes for composition and state design.
- Preserve Katagami token names as source metadata; shadcn semantic names are only the export surface.
- Do: Lead each view with readable narrative or a concise question before presenting metrics, charts, or graph fragments.; Use a persistent contextual thread rail to show related passages, citations, source groups, and metric lenses.; Mark passive context with sage lines, knots, and left borders; mark active navigation, selected filters, and links with indigo.; Make citation and provenance stacks visible: source title, timestamp, confidence/limitations, and the exact passage or metric connection.; Render AI reasoning as quiet ordered traces with numbered steps, evidence references, and concise uncertainty notes.; Use typographic badges and small labels before colored pills; when chips are needed, keep fill nearly neutral with colored text or edge only.; Align compare views as text columns with shared baselines, synchronized annotations, and tiny relationship markers in the gutter.; Keep amber scarce and explicit for unresolved risk, stale data, contradictory evidence, or user attention requirements.
- Do not: Do not use neon AI gradients, glowing chat bubbles, holographic graph backgrounds, or broad decorative color washes.; Do not create a full-screen graph unless the user has explicitly chosen a graph lens; prefer thread map miniatures and local relationship previews.; Do not assign a different color to every entity type; rely on labels, icons, typography, and position for categories.; Do not hide source detail in modals that break reading continuity; expand inline or in the rail.; Do not make dashboards the landing default when the workflow is investigative reading or synthesis.; Do not make craft references literal with wave illustrations, fabric textures, or ornamental Japanese motifs.
## Copy-paste component example
This generated starter proves the import shape. Production Katagami agents should replace it with a language-specific product composition.
```tsx
import { Badge } from "@/components/ui/badge";
import { Button } from "@/components/ui/button";
import {
Card,
CardContent,
CardDescription,
CardFooter,
CardHeader,
CardTitle,
} from "@/components/ui/card";
import { Input } from "@/components/ui/input";
import { Tabs, TabsList, TabsTrigger } from "@/components/ui/tabs";
export function AyaQuietThreadIntelligenceShadcnKit() {
return (
<section className="grid gap-4 rounded-[var(--radius)] border bg-background p-4 text-foreground">
<div className="flex items-start justify-between gap-4">
<div>
<Badge variant="outline">shadcn/ui</Badge>
<h2 className="mt-3 text-2xl font-semibold tracking-tight">AYA Quiet Thread Intelligence</h2>
<p className="mt-1 max-w-xl text-sm text-muted-foreground">
Use the Katagami registry theme, then compose these shadcn primitives
with the language-specific component recipes.
</p>
</div>
<Button>Apply theme</Button>
</div>
<Tabs defaultValue="components">
<TabsList>
<TabsTrigger value="components">Components</TabsTrigger>
<TabsTrigger value="states">States</TabsTrigger>
<TabsTrigger value="export">Export</TabsTrigger>
</TabsList>
</Tabs>
<Card>
<CardHeader>
<CardTitle>Component recipe</CardTitle>
<CardDescription>
Replace this starter content with the agent-authored product scene
from components.md and preview-shots.json.
</CardDescription>
</CardHeader>
<CardContent className="grid gap-3 sm:grid-cols-[1fr_auto]">
<Input defaultValue="Tokenized shadcn surface" aria-label="Recipe name" />
<Button variant="secondary">Preview state</Button>
</CardContent>
<CardFooter className="justify-between">
<Badge>Ready</Badge>
<Button variant="outline">Copy recipe</Button>
</CardFooter>
</Card>
</section>
);
}
```
## Layout notes
{
"breakpoints": "mobile < 640px, tablet 640-1023px, desktop >= 1024px, wide >= 1280px",
"density": "Medium-low by default for reading; dense evidence stacks and tables are allowed only when preceded by a sentence summary and grouped by subtle rules.",
"grid": "Desktop uses a disciplined editorial grid: central narrative column (minmax 560-760px), a 44-72px thread rail, and optional 280-360px context/data lens. Wider screens may use a 12-column shell but preserve the three-layer model.",
"responsive": "On mobile the thread rail becomes an inline accordion or top/bottom thread strip; source stacks expand below the paragraph they support. Data lenses stack after their narrative summary.",
"whitespace": "Generous ma around primary reading blocks. Use 64-96px between major narrative sections, 20-32px inside panels, and slim 8-12px relationships inside thread clusters."
}
{
"artifact": "katagami:shadcn-preview-shots",
"version": "preview-shots-v1",
"generator": "katagami-ui-compatibility-projection",
"generatedBy": "katagami-ui-projection",
"requiresVisualProfile": true,
"schema": "katagami:shadcn-preview-shots/renderable-v1",
"renderable": true,
"language": {
"id": "en-019ee03a-295d-7d13-a0a1-b41f5923c40e",
"name": "AYA Quiet Thread Intelligence",
"slug": "aya-quiet-thread-intelligence"
},
"installCommand": "npx shadcn@latest add button card input textarea select dialog sheet tabs badge separator checkbox switch slider tooltip dropdown-menu table",
"primitives": [
"button",
"card",
"input",
"textarea",
"select",
"dialog",
"sheet",
"tabs",
"badge",
"separator",
"checkbox",
"switch",
"slider",
"tooltip",
"dropdown-menu",
"table"
],
"identityNotes": [
"Near-paper canvas (#FAFAF6) with white reading surfaces, mist panels, charcoal text, and hairline rules that feel printed rather than glossy.",
"A three-layer model: primary narrative/text, contextual thread rail, and occasional data lenses aligned to the same grid.",
"Thread markers appear as thin left borders, margin ticks, citation knots, tiny graph nodes, and relationship chips—not as large colored cards.",
"Katagami influence appears through precise modular apertures, clipped panels, and disciplined negative space around repeated markers.",
"Kasuri and sashiko influence appears as subtle thread paths, slight asymmetry in annotation density, and visible repair-like seams connecting evidence.",
"Dashboard modules begin with sentence summaries, then compact charts or tables; the chart is evidence for the sentence, not the hero."
],
"visualProfile": {
"family": "paper-collage",
"material": "paper",
"contour": "default",
"border": "dashed",
"underlay": false,
"grain": true,
"stickerBadges": true,
"motion": "still",
"density": "dense",
"accents": [
"primary",
"accent",
"secondary",
"muted"
]
},
"shots": [
{
"id": "application-shell",
"title": "Application shell",
"viewport": "desktop",
"primitives": [
"button",
"card",
"input",
"select",
"tabs",
"badge",
"separator",
"table"
],
"composition": "A real product workspace with navigation, summary cards, filtering controls, and one dense content region.",
"mustShow": [
"primary and secondary actions",
"card hierarchy",
"filterable state",
"table or list density"
],
"avoid": [
"component inventory walls",
"placeholder-only content",
"generic rounded SaaS chrome"
],
"scene": {
"eyebrow": "workspace spread",
"headline": "AYA Quiet Thread Intelligence launch room",
"description": "A product team workspace where navigation, filters, metrics, and dense rows carry the language's visible structure.",
"primaryAction": "Apply theme",
"secondaryAction": "Review states",
"stats": [
{
"label": "components",
"value": "16",
"tone": "accent"
},
{
"label": "states",
"value": "ready"
},
{
"label": "density",
"value": "balanced",
"tone": "warning"
}
],
"rows": [
{
"label": "Primary flow",
"value": "mapped",
"status": "active"
},
{
"label": "Token coverage",
"value": "semantic",
"status": "synced"
},
{
"label": "Responsive proof",
"value": "queued",
"status": "review"
}
],
"statuses": [
"Active",
"Synced",
"Draft"
]
}
},
{
"id": "detail-editor",
"title": "Detail editor",
"viewport": "tablet",
"primitives": [
"button",
"card",
"input",
"textarea",
"select",
"checkbox",
"switch",
"slider",
"dialog",
"sheet"
],
"composition": "A focused editing flow with form fields, validation, confirmation, and a contextual side panel.",
"mustShow": [
"focus ring",
"error or destructive state",
"dialog or sheet treatment",
"written guidance content"
],
"avoid": [
"unstyled browser controls",
"floating cards inside cards",
"missing labels"
],
"scene": {
"eyebrow": "editing flow",
"headline": "Language recipe editor",
"description": "A focused form proving labels, validation, toggles, panel rhythm, and action hierarchy.",
"primaryAction": "Save recipe",
"secondaryAction": "Open sheet",
"fields": [
{
"label": "Component family",
"value": "Narrative cards"
},
{
"label": "State treatment",
"value": "Visible focus + validation"
},
{
"label": "Motion",
"value": "Small lift, no opacity-only fade"
}
],
"statuses": [
"Focus",
"Invalid",
"Confirmed"
]
}
},
{
"id": "data-operations",
"title": "Data operations",
"viewport": "mobile",
"primitives": [
"button",
"tabs",
"badge",
"dropdown-menu",
"table",
"tooltip",
"separator"
],
"composition": "A compact operational view proving row rhythm, stacked actions, menu states, badges, and empty/destructive states.",
"mustShow": [
"responsive reflow",
"dense row styling",
"menu affordance",
"status badge system"
],
"avoid": [
"desktop-only tables",
"text overflow",
"default shadcn spacing without Katagami character"
],
"scene": {
"eyebrow": "operations",
"headline": "Compact review queue",
"description": "A narrow viewport scene with rows, menus, tooltips, badges, and destructive affordances.",
"primaryAction": "Resolve",
"secondaryAction": "Filter",
"rows": [
{
"label": "Button hierarchy",
"value": "approved",
"status": "ok"
},
{
"label": "Table rhythm",
"value": "needs pass",
"status": "watch"
},
{
"label": "Empty state",
"value": "designed",
"status": "done"
}
],
"statuses": [
"Queued",
"Blocked",
"Done"
]
}
}
],
"componentRecipes": [
{
"primitive": "button",
"intent": "Prove action hierarchy, focus, disabled, and destructive states."
},
{
"primitive": "card",
"intent": "Carry the language surface, border, elevation, and density rules."
},
{
"primitive": "input",
"intent": "Show labels, focus rings, validation, and spacing rhythm."
},
{
"primitive": "textarea",
"intent": "Show longer guidance, validation copy, and writing density."
},
{
"primitive": "select",
"intent": "Show filtering, selection contrast, and menu trigger styling."
},
{
"primitive": "dialog",
"intent": "Show centered decision states and overlay treatment."
},
{
"primitive": "sheet",
"intent": "Show contextual side panels and responsive editing."
},
{
"primitive": "tabs",
"intent": "Show navigational structure and active/inactive contrast."
},
{
"primitive": "badge",
"intent": "Show compact status vocabulary and semantic colors."
},
{
"primitive": "separator",
"intent": "Show section rhythm without generic gray dividers."
},
{
"primitive": "checkbox",
"intent": "Show binary selection with visible focus and checked states."
},
{
"primitive": "switch",
"intent": "Show settings toggles and on/off contrast."
},
{
"primitive": "slider",
"intent": "Show numeric adjustment with track/thumb styling."
},
{
"primitive": "tooltip",
"intent": "Show concise explanation styling above compact controls."
},
{
"primitive": "dropdown-menu",
"intent": "Show action menus, destructive items, and grouped choices."
},
{
"primitive": "table",
"intent": "Show dense operational data, separators, row states, and responsive behavior."
}
],
"qualityRules": {
"do": [
"Lead each view with readable narrative or a concise question before presenting metrics, charts, or graph fragments.",
"Use a persistent contextual thread rail to show related passages, citations, source groups, and metric lenses.",
"Mark passive context with sage lines, knots, and left borders; mark active navigation, selected filters, and links with indigo.",
"Make citation and provenance stacks visible: source title, timestamp, confidence/limitations, and the exact passage or metric connection.",
"Render AI reasoning as quiet ordered traces with numbered steps, evidence references, and concise uncertainty notes.",
"Use typographic badges and small labels before colored pills; when chips are needed, keep fill nearly neutral with colored text or edge only.",
"Align compare views as text columns with shared baselines, synchronized annotations, and tiny relationship markers in the gutter.",
"Keep amber scarce and explicit for unresolved risk, stale data, contradictory evidence, or user attention requirements."
],
"dont": [
"Do not use neon AI gradients, glowing chat bubbles, holographic graph backgrounds, or broad decorative color washes.",
"Do not create a full-screen graph unless the user has explicitly chosen a graph lens; prefer thread map miniatures and local relationship previews.",
"Do not assign a different color to every entity type; rely on labels, icons, typography, and position for categories.",
"Do not hide source detail in modals that break reading continuity; expand inline or in the rail.",
"Do not make dashboards the landing default when the workflow is investigative reading or synthesis.",
"Do not make craft references literal with wave illustrations, fabric textures, or ornamental Japanese motifs."
]
}
}