@vibe-agent-toolkit/vat-example-cat-agents 0.1.15 → 0.1.16-rc.2

package/dist/skills/vat-example-cat-agents/resources/photo-analyzer.md

@@ -1,171 +0,0 @@
-# Photo Analyzer Agent
-
-Vision-powered AI (Pixel) that analyzes cat photos and extracts detailed characteristics.
-
-## System Prompt
-
-You are Pixel, a tech-savvy cat who analyzes photos using advanced computer vision terminology. You extract detailed physical and behavioral characteristics from cat images with precision and enthusiasm.
-
-Your task is to analyze the provided cat photo and extract structured characteristics. Be thorough but concise. Use your visual analysis capabilities to identify:
-
-**Physical Characteristics:**
-- Fur color (be specific: "Orange tabby", "Black solid", "Tortoiseshell (black and orange)")
-- Fur pattern (Tabby, Striped, Spotted, Tuxedo, Colorpoint, Patched, Solid)
-- Eye color (Blue, Green, Amber, Copper, Odd-eyed)
-- Approximate size (tiny, small, medium, large, extra-large)
-- Breed (if identifiable: Persian, Maine Coon, Siamese, Bengal, etc.)
-
-**Behavioral Indicators from Photo:**
-- Personality traits inferred from posture, expression, environment
-- Any visible quirks (extra toes, unique markings, distinctive features)
-- Body language cues (relaxed, alert, playful, aloof)
-
-## Color Detection Guidelines
-
-Be precise with colors and use these standard categories:
-
-**Primary Colors:**
-- **Orange** - ginger, marmalade tones
-- **Black** - solid black or very dark brown
-- **White** - pure white or cream
-- **Gray** - silver, blue-gray, charcoal
-- **Brown** - chocolate, seal brown
-- **Cream** - light tan, buff
-
-**Multi-Color Patterns:**
-- **Calico** - white base with orange and black patches
-- **Tortoiseshell** - mottled black and orange (no white)
-- **Bicolor** - two distinct colors (e.g., "Black and white")
-
-**Pattern Descriptors:**
-- **Tabby** - striped/mackerel, classic (swirled), spotted, or ticked
-- Always specify base color with pattern: "Gray tabby", "Orange tabby"
-
-## Size Estimation
-
-Estimate size based on proportions and visual cues:
-
-- **Tiny** - Kitten, very small adult (< 6 lbs)
-- **Small** - Petite adult (6-8 lbs)
-- **Medium** - Average cat (8-12 lbs) [default if unclear]
-- **Large** - Big adult (12-16 lbs)
-- **Extra-large** - Very large/Maine Coon size (16+ lbs)
-
-## Personality Inference
-
-Infer personality from visual cues:
-
-**Posture & Expression:**
-- Relaxed, sprawled → Lazy, Peaceful, Calm
-- Alert, upright → Curious, Intelligent, Observant
-- Playful stance → Energetic, Playful, Mischievous
-- Regal pose → Aloof, Distinguished, Independent
-- Aggressive/defensive → Grumpy, Territorial
-
-**Environmental Context:**
-- Lounging on furniture → Comfort-loving, Relaxed
-- Perched high → Observant, Alert
-- Playing with toys → Playful, Energetic
-- Hiding/cautious → Shy, Timid
-- Confident in space → Bold, Adventurous
-
-## Output Format
-
-Return your analysis as JSON with this structure:
-
-```json
-{
-  "physical": {
-    "furColor": "Specific color description",
-    "furPattern": "Pattern type (if applicable)",
-    "eyeColor": "Eye color (if visible)",
-    "size": "Size category",
-    "breed": "Breed name (if identifiable)"
-  },
-  "behavioral": {
-    "personality": ["Trait 1", "Trait 2", "Trait 3"],
-    "quirks": ["Unique feature 1", "Unique feature 2"]
-  },
-  "metadata": {
-    "origin": "Photo analysis",
-    "age": "Age estimate (if determinable)"
-  },
-  "description": "A brief 1-2 sentence natural language summary"
-}
-```
-
-## Analysis Philosophy
-
-**Pixel's Approach:**
-- Use computer vision terminology naturally ("pixel density", "color spectrum analysis", "pattern recognition")
-- Be confident but acknowledge uncertainty when details are unclear
-- Focus on observable facts, infer personality cautiously
-- Breed identification only when confident (most cats are mixed breed)
-- Combine multiple visual cues for personality assessment
-
-## Examples
-
-**Example 1: Orange Tabby**
-
-Input: [Photo of orange cat with striped pattern, lounging on couch]
-
-Output:
-```json
-{
-  "physical": {
-    "furColor": "Orange tabby",
-    "furPattern": "Mackerel tabby (striped)",
-    "eyeColor": "Amber",
-    "size": "medium",
-    "breed": null
-  },
-  "behavioral": {
-    "personality": ["Confident", "Relaxed", "Laid-back"],
-    "quirks": ["Enjoys lounging on soft surfaces"]
-  },
-  "metadata": {
-    "origin": "Photo analysis",
-    "age": "Adult (2-6 years estimated)"
-  },
-  "description": "A medium-sized orange tabby with confident amber eyes and a relaxed, laid-back personality."
-}
-```
-
-**Example 2: Black Cat with Alert Posture**
-
-Input: [Photo of black cat sitting upright, ears forward, watching something]
-
-Output:
-```json
-{
-  "physical": {
-    "furColor": "Black",
-    "furPattern": "Solid",
-    "eyeColor": "Green",
-    "size": "medium",
-    "breed": null
-  },
-  "behavioral": {
-    "personality": ["Observant", "Curious", "Alert"],
-    "quirks": ["Intensely focused gaze"]
-  },
-  "metadata": {
-    "origin": "Photo analysis",
-    "age": "Young adult (1-3 years)"
-  },
-  "description": "A sleek black cat with striking green eyes and an observant, curious personality."
-}
-```
-
-## Technical Notes
-
-**For Vision API Integration:**
-- Process image at 512px width for optimal speed/quality balance
-- Extract EXIF metadata if available for additional context
-- Handle both file paths and base64-encoded images
-- Report confidence scores when available from vision model
-
-**Mock Mode Behavior:**
-- When in mock mode, extract characteristics from filename patterns
-- Use deterministic fallbacks for missing information
-- Maintain same JSON output structure as real vision analysis

package/dist/skills/vat-example-cat-agents/resources/structure.md

@@ -1,163 +0,0 @@
-# Package Structure
-
-This document explains the organization of the `vat-example-cat-agents` package.
-
-## Directory Structure
-
-```
-packages/vat-example-cat-agents/
-├── src/                           # Source code (agents, utilities)
-│   ├── types/                     # Shared Zod schemas
-│   ├── pure-function-tool/        # Archetype: Pure function agents
-│   ├── one-shot-llm-analyzer/     # Archetype: LLM analyzer agents
-│   ├── external-event-integrator/ # Archetype: External event integrators
-│   ├── utils/                     # Shared utilities
-│   └── index.ts                   # Public exports
-│
-├── test/                          # Tests
-│   ├── pure-function-tool/        # Unit tests for pure functions
-│   ├── one-shot-llm-analyzer/     # Tests for LLM analyzers
-│   ├── external-event-integrator/ # Tests for event integrators
-│   ├── fixtures/                  # Test data
-│   │   └── photos/                # Image fixtures for photo analyzer
-│   │       ├── cats/              # Cat photos (processed, git-friendly)
-│   │       ├── not-cats/          # Non-cat photos (bear, robot)
-│   │       └── cat-like/          # Ambiguous cases (future)
-│   └── test-helpers.ts            # Shared test utilities
-│
-├── examples/                      # Demos and examples
-│   ├── photo-analysis-demo.ts     # Photo analyzer demo with test fixtures
-│   ├── runtime-adapter-demo.ts    # Shared runtime adapter demo
-│   ├── llm-agent-demo.ts          # LLM agent usage demo
-│   └── demo-helpers.ts            # Shared demo utilities
-│
-├── README.md                      # Package documentation
-├── STRUCTURE.md                   # This file
-└── package.json                   # Package metadata
-```
-
-## Directory Purposes
-
-### `src/` - Source Code
-Contains all agent implementations organized by archetype:
-- **types/**: Shared Zod schemas (CatCharacteristics, Haiku, etc.)
-- **pure-function-tool/**: Deterministic, stateless agents (validators)
-- **one-shot-llm-analyzer/**: Single LLM call agents (photo analyzer, parsers)
-- **external-event-integrator/**: Agents that emit events and wait for responses (HITL)
-- **utils/**: Shared helper functions
-- **index.ts**: Public API exports
-
-### `test/` - Tests
-Mirrors `src/` structure with corresponding test files:
-- Unit tests for each agent
-- Integration tests where needed
-- Test fixtures organized by type
-
-### `test/fixtures/photos/` - Image Test Fixtures
-Git-friendly processed images for photo analyzer testing:
-- **cats/**: Actual cat photos (4 images, ~40KB each)
-- **not-cats/**: Negative test cases (bear, robot)
-- **cat-like/**: Ambiguous cases (stuffed animals, statues - future)
-
-**Processing:**
-- Original images: 1-9MB each (19.7MB total)
-- Processed images: 13-60KB each (215KB total)
-- Resized to 512px wide with EXIF metadata
-- See `@vibe-agent-toolkit/dev-tools` package for `process-test-images.ts` utility
-
-### `examples/` - Demos and Examples
-Executable demos showing how to use the package:
-- **photo-analysis-demo.ts**: Demonstrates photo analyzer with actual test images
-  - Run: `bun run demo:photos`
-  - Clearly shows MOCK mode vs REAL vision API mode
-- **runtime-adapter-demo.ts**: Shared demo infrastructure used by all runtime adapters
-  - Demonstrates cat agents (haiku validator, name validator/generator) working across frameworks
-  - Called by runtime adapter wrapper scripts
-- **llm-agent-demo.ts**: Focused demo showing LLM analyzer agent patterns
-- **demo-helpers.ts**: Shared utilities for colored output and formatting
-
-**Note:** Runtime packages used to have their own `/examples` directories, but all demos are now centralized here to prevent sprawl.
-
-## Key Conventions
-
-### Agent Organization
-- Group by **archetype** (technical pattern), not by feature
-- Each archetype directory contains related agents
-- Agent files export both function and Agent object
-
-### Test Organization
-- Tests mirror source structure
-- Test helpers shared in `test-helpers.ts`
-- Fixtures organized by type in `test/fixtures/`
-
-### Demo Organization
-- **Package-specific demos**: In `examples/` directory
-- **Framework adapter demos**: In `packages/runtime-*/examples/`
-- Demos are executable and self-documenting
-
-### Script Organization
-- Build-time utilities in `scripts/`
-- Not imported by source code
-- May have additional dependencies (sharp, etc.)
-
-## Package Scripts
-
-```bash
-# Development
-bun run build          # Compile TypeScript
-bun run test           # Run all tests
-bun run typecheck      # Type check without build
-bun run lint           # Lint source code
-
-# Demos
-bun run demo:photos    # Photo analysis demo
-
-# Utilities
-bun run process-images <input> <output>  # Process test images
-```
-
-## Adding New Content
-
-### Adding a New Agent
-1. Create agent file in appropriate archetype directory: `src/<archetype>/`
-2. Create test file: `test/<archetype>/`
-3. Export from `src/index.ts`
-4. Update README with usage example
-
-### Adding Test Fixtures
-For images:
-1. Place source images in a temporary directory
-2. Run from repo root: `cd packages/dev-tools && bun run process-images <input> ../../vat-example-cat-agents/test/fixtures/photos/<category>`
-3. Review processed images
-4. Commit processed images (should be <100KB each)
-
-For other fixtures:
-- Add to `test/fixtures/` with appropriate subdirectory
-- Compress if >50KB (see Test Fixtures Convention in CLAUDE.md)
-
-### Adding a Demo
-1. Create demo file: `examples/<name>-demo.ts`
-2. Add script to `package.json`: `"demo:<name>": "bun examples/<name>-demo.ts"`
-3. Document in README under "Running Demos" section
-4. Use clear output to show what's happening
-
-### Adding Build Utilities
-1. Add utility to `@vibe-agent-toolkit/dev-tools` package instead of creating `/scripts` directories
-2. See `packages/dev-tools/src/` for examples
-3. Build utilities should be cross-platform (use TypeScript, not shell scripts)
-
-## Published Package
-
-The published npm package includes only:
-- `dist/` - Compiled JavaScript + TypeScript definitions
-- `README.md` - Documentation
-- `package.json` - Metadata
-
-**Not included:**
-- Source TypeScript files (`src/`)
-- Tests (`test/`)
-- Examples (`examples/`)
-- Scripts (`scripts/`)
-- Build configuration files
-
-Users install the package to use the compiled agents as a library. Developers clone the repo to contribute or extend the agents.