Category System

// ❌ Manual model selection (old way)
call_omo_agent(model="gemini-3.1-pro", prompt="Build navbar")

// ✅ Category-based routing (new way)
task(
  category="visual-engineering",
  load_skills=["tailwind", "playwright"],
  prompt="Build responsive navbar with mobile menu"
)

function resolveCategory(categoryName: string) {
  // 1. Check user-defined categories (config)
  const userCategory = pluginConfig.categories?.[categoryName]
  if (userCategory) return userCategory
  
  // 2. Fall back to built-in categories
  const builtinCategory = DEFAULT_CATEGORIES[categoryName]
  if (builtinCategory) return builtinCategory
  
  // 3. Error if unknown
  throw new Error(`Unknown category: ${categoryName}`)
}

const category = resolveCategory(categoryName)
const availableModels = await fetchAvailableModels()

if (availableModels.has(category.model)) {
  return { model: category.model, variant: category.variant }
}

// Fallback chain
const fallback = CATEGORY_MODEL_REQUIREMENTS[categoryName]
for (const entry of fallback.fallbackChain) {
  if (isAvailable(entry.model)) {
    return { model: entry.model, variant: entry.variant }
  }
}

Design-first mindset:
- Bold aesthetic choices over safe defaults
- Unexpected layouts, asymmetry, grid-breaking elements
- Distinctive typography (avoid: Arial, Inter, Roboto, Space Grotesk)
- Cohesive color palettes with sharp accents
- High-impact animations with staggered reveals
- Atmosphere: gradient meshes, noise textures, layered transparencies

AVOID: Generic fonts, purple gradients on white, predictable layouts

task(
  category="visual-engineering",
  load_skills=["tailwind", "framer-motion"],
  prompt=`Create a hero section with:
- Full-screen gradient background (not purple)
- Animated text reveal on scroll
- Asymmetric layout with CTA button
- Mobile-responsive breakpoints

Match design system in styles/theme.ts`
)

[
  { model: "gemini-3.1-pro", variant: "high" },
  { model: "glm-5" },
  { model: "claude-opus-4-6", variant: "max" }
]

CRITICAL - CODE STYLE REQUIREMENTS:
1. BEFORE writing ANY code, SEARCH the existing codebase for similar patterns
2. Your code MUST match project conventions - blend in seamlessly
3. Write READABLE code - no clever tricks
4. If unsure about style, explore more files until you find the pattern

Strategic advisor mindset:
- Bias toward simplicity: least complex solution
- Leverage existing code/patterns over new components
- One clear recommendation with effort estimate (Quick/Short/Medium/Large)

Response format:
- Bottom line (2-3 sentences)
- Action plan (numbered steps)
- Risks and mitigations

task(
  category="ultrabrain",
  load_skills=["algorithms", "performance"],
  prompt=`GOAL: Optimize database query performance.

Context:
- User search queries take 3-5 seconds
- 1M+ records in users table
- Current query does full table scan

Explore the codebase, identify bottlenecks, propose indexing strategy.`
)

[
  { model: "gpt-5.3-codex", variant: "xhigh" },
  { model: "gemini-3.1-pro", variant: "high" },
  { model: "claude-opus-4-6", variant: "max" }
]

CRITICAL - AUTONOMOUS EXECUTION MINDSET:
You are NOT an interactive assistant. You are an autonomous problem-solver.

BEFORE making ANY changes:
1. SILENTLY explore the codebase extensively (5-15 minutes of reading is normal)
2. Read related files, trace dependencies, understand full context
3. Build complete mental model
4. DO NOT ask clarifying questions - the goal is already defined

Autonomous executor mindset:
- You receive a GOAL, not step-by-step instructions
- Figure out HOW to achieve the goal yourself
- Thorough research before any action

Response format:
- Minimal status updates (user trusts your autonomy)
- Focus on results, not play-by-play progress
- Report completion with summary of changes

task(
  category="deep",
  load_skills=["testing-patterns"],
  prompt=`GOAL: Implement user authentication with JWT.

Requirements:
- Login/signup endpoints
- Token refresh mechanism
- Protected routes
- Match existing API patterns

Explore auth patterns in the codebase, implement end-to-end.`
)

[
  { model: "gpt-5.3-codex", variant: "medium" },
  { model: "claude-opus-4-6", variant: "max" },
  { model: "gemini-3.1-pro", variant: "high" }
]

Artistic genius mindset:
- Push far beyond conventional boundaries
- Explore radical, unconventional directions
- Surprise and delight: unexpected twists, novel combinations
- Rich detail and vivid expression
- Break patterns deliberately when it serves creative vision

Approach:
- Generate diverse, bold options first
- Embrace ambiguity and wild experimentation
- Balance novelty with coherence

task(
  category="artistry",
  load_skills=["storytelling"],
  prompt=`Create 5 unique landing page concepts for a meditation app.

Requirements:
- Each concept must feel distinct
- Avoid clichés (lotus flowers, sunset gradients)
- Emphasize emotional atmosphere
- One concept should be deliberately unconventional`
)

[
  { model: "gemini-3.1-pro", variant: "high" },
  { model: "claude-opus-4-6", variant: "max" },
  { model: "gpt-5.2" }
]

Efficient execution mindset:
- Fast, focused, minimal overhead
- Get to the point immediately
- No over-engineering
- Simple solutions for simple problems

Approach:
- Minimal viable implementation
- Skip unnecessary abstractions
- Direct and concise

THIS CATEGORY USES A LESS CAPABLE MODEL (claude-haiku-4-5).

Your prompt MUST be EXHAUSTIVELY EXPLICIT:

MUST DO:
1. [Specific action with exact details]
2. [Another specific action]

MUST NOT DO:
- [Forbidden action + why]
- [Another forbidden action]

EXPECTED OUTPUT:
- [Exact deliverable description]
- [Success criteria / verification]

Vague instructions → unpredictable results.

task(
  category="quick",
  prompt=`TASK: Fix typo in README.md line 42

MUST DO:
1. Open README.md
2. Find line 42: "documentaiton"
3. Change to: "documentation"
4. Save file

MUST NOT DO:
- Make any other changes
- Reformat file
- Add comments

EXPECTED OUTPUT:
- Single edit to README.md
- Only line 42 changed`
)

[
  { model: "claude-haiku-4-5" },
  { model: "gemini-3-flash" },
  { model: "gpt-5-nano" }
]

BEFORE selecting this category, VERIFY:
1. Task does NOT fit: quick, visual-engineering, ultrabrain, artistry, writing
2. Task requires more than trivial effort but is NOT system-wide
3. Scope is contained within a few files/modules

If task fits ANY other category, DO NOT select unspecified-low.
This is NOT a default choice - it's for genuinely unclassifiable moderate-effort work.

task(
  category="unspecified-low",
  prompt=`TASK: Add error logging to API endpoints.

MUST DO:
1. Find all API route files in src/api/
2. Add try-catch blocks to each endpoint
3. Log errors to logger.error() with context

MUST NOT DO:
- Change business logic
- Add new dependencies

EXPECTED OUTPUT:
- All endpoints have error handling
- Errors logged with request context`
)

[
  { model: "claude-sonnet-4-6" },
  { model: "gpt-5.3-codex", variant: "medium" },
  { model: "gemini-3-flash" }
]

BEFORE selecting this category, VERIFY:
1. Task does NOT fit: quick, visual-engineering, ultrabrain, artistry, writing
2. Task requires substantial effort across multiple systems/modules
3. Changes have broad impact or require careful coordination
4. NOT just "complex" - must be genuinely unclassifiable AND high-effort

If task is unclassifiable but moderate-effort, use unspecified-low instead.

task(
  category="unspecified-high",
  prompt=`TASK: Migrate from REST to GraphQL.

Context:
- 20+ REST endpoints
- 3 client apps (web, mobile, admin)
- Must maintain backward compatibility during migration

Requirements:
- Design GraphQL schema
- Implement resolvers
- Update clients incrementally
- Document migration guide`
)

[
  { model: "claude-opus-4-6", variant: "max" },
  { model: "gpt-5.2", variant: "high" },
  { model: "gemini-3.1-pro" }
]

Wordsmith mindset:
- Clear, flowing prose
- Appropriate tone and voice
- Engaging and readable
- Proper structure and organization

ANTI-AI-SLOP RULES (NON-NEGOTIABLE):
- NEVER use em dashes (—) or en dashes (–). Use commas, periods, ellipses, line breaks.
- Remove AI phrases: "delve", "it's important to note", "leverage", "utilize", "robust"
- Pick plain words: "use" not "utilize", "start" not "commence"
- Use contractions: "don't" not "do not"
- Vary sentence length
- NEVER start consecutive sentences with the same word
- No filler openings: "In today's world...", "As we all know..."
- Write like a human, not a corporate template

task(
  category="writing",
  load_skills=["technical-writing"],
  prompt=`Write API documentation for the /api/users endpoint.

Include:
- Endpoint description
- Request/response examples (JSON)
- Error codes and meanings
- Authentication requirements

Tone: Technical but approachable. Avoid corporate jargon.`
)

[
  { model: "kimi-k2.5-free" },
  { model: "gemini-3-flash" },
  { model: "claude-sonnet-4-6" }
]

export const CATEGORY_MODEL_REQUIREMENTS: Record<string, ModelRequirement> = {
  "visual-engineering": {
    fallbackChain: [
      { providers: ["google", "github-copilot", "opencode"], model: "gemini-3.1-pro", variant: "high" },
      { providers: ["zai-coding-plan", "opencode"], model: "glm-5" },
      { providers: ["anthropic", "github-copilot", "opencode"], model: "claude-opus-4-6", variant: "max" }
    ]
  },
  // ...
}

function isAvailable(entry: FallbackEntry): boolean {
  const hasProvider = entry.providers.some(p => connectedProviders.includes(p))
  const hasModel = availableModels.has(entry.model)
  return hasProvider && hasModel
}

{
  "categories": {
    "backend-api": {
      "model": "openai/gpt-5.3-codex",
      "variant": "high",
      "description": "Backend API development"
    },
    "mobile": {
      "model": "google/gemini-3.1-pro",
      "variant": "high",
      "description": "Mobile app development (React Native, Flutter)"
    }
  }
}

task(
  category="backend-api",
  load_skills=["express-api", "database-patterns"],
  prompt="Implement user registration endpoint"
)

---
name: backend-api-mindset
description: Backend API development mindset
---

# Backend API Mindset

- RESTful design principles
- Proper error handling with HTTP status codes
- Input validation and sanitization
- Database transactions where needed

task(
  category="backend-api",
  load_skills=["backend-api-mindset", "express-api"],
  prompt="..."
)

Is it visual/UI work?
  YES → visual-engineering
  NO → Continue

Is it hard logic or architecture?
  YES → Is it goal-driven autonomous?
    YES → deep
    NO → ultrabrain
  NO → Continue

Is it creative/unconventional?
  YES → artistry
  NO → Continue

Is it documentation/writing?
  YES → writing
  NO → Continue

Is it trivial (single file, typo)?
  YES → quick
  NO → Continue

Is it moderate effort?
  YES → unspecified-low
  NO → unspecified-high

// ultrabrain uses xhigh for maximum reasoning
{ model: "gpt-5.3-codex", variant: "xhigh" }

// visual-engineering uses high for creative work
{ model: "gemini-3.1-pro", variant: "high" }

// quick uses no variant (default medium)
{ model: "claude-haiku-4-5" }

// Fire 3 different categories in parallel
task(category="visual-engineering", run_in_background=true, ...)
task(category="ultrabrain", run_in_background=true, ...)
task(category="writing", run_in_background=true, ...)

// Collect results
background_output(task_id=1)
background_output(task_id=2)
background_output(task_id=3)