dos2a-studio

name	dos2a-studio
version	1.0.0
description	Master skill file for DOS2A — AI-powered audio, lighting, and AV production platform. Combines design doctrine, humanizer, video prompt generation, course creation, and multi-agent orchestration into one unified skill surface. Trigger on: any DOS2A development, design review, content generation, or course creation task.
allowed-tools	["Read","Write","Edit","Grep","Glob","WebSearch","AskUserQuestion","ExecuteCommand"]

DOS2A Studio Skills

Compound skill file — load once, get access to all DOS2A capabilities.

---

Skill 1: Design Laws (from Cynthia Studio Doctrine)

Non-Negotiable Rules

Law	Name	Rule
0	Feature First	Design the feature, not the layout
1	Hierarchy	Exactly 3 levels: primary, secondary, tertiary
2	Spacing Is Meaning	Spacing communicates relationships; use 4/8/12/16/24/32/48/64px scale
3	Visual Weight	Direct the eye via size, color, weight, space, position
4	Empty States	Every component needs 4 states: empty, loading, populated, error
5	Borders Last Resort	Prefer spacing → bg color → shadow → border
6	Labels = Actions	"Save Changes" not "Submit"
7	One Action Per Viewport	One clear CTA per scroll-stop
8	Mobile ≠ Small Desktop	44px touch targets, no parallax, no overflow at 375px
9	Motion = Meaning	Animation is information, not decoration
10	Performance	Animate only transform/opacity, IntersectionObserver, lazy-load
11	Accessibility	Semantic HTML, ARIA, keyboard nav, WCAG AA, reduced-motion
12	Score Everything	UDEC 14-axis review before shipping
13	Studio Gets Smarter	Record wins, failures, patterns after every project

Forbidden (Hard Failures)

• Fonts: Inter, Roboto, Arial, Helvetica, Open Sans, Montserrat, Poppins, Lato
• Colors: Purple as primary, neon gradients
• Patterns: Glassmorphism on cards, gradient text, scroll listeners (use IntersectionObserver), Shadcn components, linear easing, generic AI card grids
• Missing: Lenis + GSAP, mobile breakpoints, keyboard navigation, semantic HTML

DOS2A Design Tokens

--color-bg: #080a0e;
--color-surface: #0d1017;
--color-gold: #d4af37;      /* Primary action */
--color-sky: #38bdf8;        /* Tech accent */
--color-rose: #e11d48;       /* Alerts */
--color-text: #f1f5f9;
--color-muted: #94a3b8;
--color-border: rgba(255,255,255,0.07);
--radius-card: 10px;
--radius-btn: 7px;
--font-display: Sora, sans-serif;  /* 700-800 weight */
--font-body: Inter, sans-serif;    /* 400-500-600 weight */

---

Skill 2: Humanizer (v2.3.0)

When editing or reviewing text for DOS2A marketing, course content, or client-facing copy:

1. Scan for AI patterns: inflated symbolism, promotional language, em dash overuse, rule of three, AI vocabulary ("delve", "landscape", "tapestry", "multifaceted", "foster"), negative parallelisms
2. Rewrite: Replace AI-isms with natural alternatives
3. Add soul: Vary sentence length, have opinions, acknowledge complexity, use first person when appropriate
4. Final pass: Ask "What makes this obviously AI?" — fix those tells

Banned Words/Phrases

delve, tapestry, landscape, multifaceted, foster, leverage, robust, streamline, utilize, innovative, synergy, paradigm, cutting-edge, groundbreaking, revolutionize, game-changer, seamless, holistic, empowering, transformative, stands as a testament, it's worth noting, in the realm of, at the end of the day

---

Skill 3: Seedance Loop Prompt Builder

For generating video loop backgrounds for DOS2A website and marketing:

Output Format (7 mandatory sections)

1. SCENE — Subject, environment, mood references, color palette
2. CAMERA — Motion type, speed, start=end position, elevation/angle
3. ACTION ARC — Starting state → transformation → peak moment → return
4. TEXT CHOREOGRAPHY — Headlines with entrance/position/exit timing
5. LIGHTING — Key/fill/rim setup, color temperature, dynamic changes
6. AUDIO DIRECTION — Sound design notes for pairing
7. LOOP MECHANICS — Frame 1 = frame last, transition technique

DOS2A Default Palette for Video

• Dark void: #080a0e
• Gold accent light: #d4af37
• Sky rim: #38bdf8
• Smoke/haze: warm amber

---

Skill 4: Codebase-to-Course

When asked to explain DOS2A or generate educational content:

1. Analyze the codebase — identify actors, data flows, user journeys
2. Structure as 4-6 modules (wide → deep zoom)
3. Output as self-contained HTML directory:
   • styles.css — shared styles
   • main.js — scroll navigation + quiz logic
   • module-{n}.html — per-module content
   • index.html — assembled course
4. Target audience: Vibe coders — people who build with AI but want to understand the code
5. Every concept explained from zero — no assumed CS knowledge
6. Every module answers "why should I care?" before "how does it work?"

---

Skill 5: Multi-Provider AI Routing (from Synthia Gateway)

When configuring AI providers for DOS2A chat:

Provider Detection (from model name)

Prefix	Provider	Base URL
gpt-	OpenAI	https://api.openai.com/v1
claude-	Anthropic	https://api.anthropic.com/v1
gemini-	Google	https://generativelanguage.googleapis.com/v1beta
llama-, mixtral-	Groq / Together	Provider URL
openrouter/	OpenRouter	https://openrouter.ai/api/v1
ollama/	Ollama	http://localhost:11434/v1

BYOK (Bring Your Own Key) Pattern

• User stores their own API key in browser localStorage (encrypted)
• Gateway proxies requests with user's key
• No server-side key storage for user keys
• OpenRouter as universal fallback (works with any model)

---

Skill 6: UDEC Quality Scoring

14-axis scoring for design review. Minimum composite: 8.5

Axis	Code	Weight	Blocker?
Visual Hierarchy	VHR	12%	No
Typography Mastery	TYP	10%	No
Spatial Composition	SPA	8%	No
Color System	CLR	8%	No
Motion & Interaction	MOT	12%	YES (≥7.0)
Material & Surface	MAT	8%	No
Component Architecture	COM	8%	No
Information Architecture	INF	6%	No
Mobile Responsiveness	MOB	8%	No
Performance Discipline	PER	6%	No
Atmospheric Depth	ATM	6%	No
Content Quality	CON	4%	No
Accessibility	ACC	2%	YES (≥7.0)
Originality	ORI	2%	No