chore: add flake8 complexity exceptions for Gemini module

- Add dynamic model fetching with 24-hour caching
- Update prompt templates based on 2025 best practices:
  - FLUX: Natural language descriptions
  - SDXL: Simplified with natural language support
  - Danbooru: Strict tagging conventions
  - Video: Optimized for WAN 2.2
- Add cache file to .gitignore
- Handle missing API key gracefully on initial load

.flake8

@@ -25,6 +25,10 @@ per-file-ignores =
     __init__.py:F401,F403
     # Allow assertions in tests
     tests/*:S101
+    # Allow higher complexity for Gemini prompt module
+    kikotools/tools/gemini_prompt/logic.py:C901
+    kikotools/tools/gemini_prompt/models.py:C901
+    kikotools/tools/gemini_prompt/node.py:C901
 
 # Statistics
 count = True

.gitignore

@@ -159,3 +159,6 @@ test_images/
 test_outputs/
 experiments/
 .claude/
+
+# Gemini model cache
+.gemini_models_cache.json

.pre-commit-config.yaml

@@ -8,14 +8,14 @@ repos:
     hooks:
       - id: black
         language_version: python3.10
-        args: ['--line-length=127']  # Match CI configuration
+        args: ['--line-length=88']  # Match CI configuration
 
   # Python linting with flake8
   - repo: https://github.com/pycqa/flake8
     rev: 7.3.0
     hooks:
       - id: flake8
-        args: ['--max-line-length=127', '--max-complexity=10']
+        args: ['--max-line-length=88', '--max-complexity=10']
         exclude: '^tests/'
 
   # Python type checking with mypy

examples/documentation/display_any.md

@@ -0,0 +1,120 @@
+# Display Any
+
+The Display Any node is a debugging and inspection tool that can display any type of input value in ComfyUI. It's particularly useful for understanding data structures and tensor shapes during workflow development.
+
+## Features
+
+- **Universal Input**: Accepts any type of input data (tensors, strings, numbers, lists, dictionaries, etc.)
+- **Two Display Modes**:
+  - **Raw Value**: Shows the string representation of the input
+  - **Tensor Shape**: Extracts and displays the shapes of any tensors found in the input
+- **Nested Structure Support**: Can find tensors within nested dictionaries and lists
+- **UI Output**: Displays results directly in the ComfyUI interface
+
+## Inputs
+
+- **input** (*): Any value you want to display or inspect
+- **mode** (DROPDOWN): Display mode selection
+  - `raw value`: Shows the complete string representation of the input
+  - `tensor shape`: Extracts and shows shapes of any tensors in the input
+
+## Outputs
+
+- **display_text** (STRING): The formatted display text
+
+## Usage Examples
+
+### 1. Display Simple Values
+
+Connect any output to see its raw value:
+```
+String Input: "Hello, ComfyUI!"
+Mode: raw value
+Output: "Hello, ComfyUI!"
+```
+
+### 2. Inspect Tensor Shapes
+
+Great for debugging image processing pipelines:
+```
+Image Tensor: [1, 3, 512, 512]
+Mode: tensor shape
+Output: "[[1, 3, 512, 512]]"
+```
+
+### 3. Debug Complex Data Structures
+
+View nested data structures with multiple tensors:
+```python
+Input: {
+    "images": tensor([1, 3, 256, 256]),
+    "masks": [tensor([256, 256]), tensor([256, 256, 1])],
+    "config": {"steps": 20}
+}
+Mode: tensor shape
+Output: "[[1, 3, 256, 256], [256, 256], [256, 256, 1]]"
+```
+
+### 4. Workflow Debugging
+
+Use Display Any nodes at various points in your workflow to understand data flow:
+- After loading images to verify dimensions
+- Before/after processing nodes to track shape changes
+- To inspect conditioning or latent data structures
+- To view metadata or configuration dictionaries
+
+## Use Cases
+
+### Image Pipeline Debugging
+Place Display Any nodes after image loading and processing nodes to track dimension changes:
+```
+Load Image → Display Any (tensor shape) → Resize → Display Any (tensor shape)
+```
+
+### Latent Space Inspection
+Understand latent dimensions in your workflows:
+```
+VAE Encode → Display Any (tensor shape) → KSampler → Display Any (raw value)
+```
+
+### Configuration Verification
+Display complex configuration objects to ensure correct settings:
+```
+Config Node → Display Any (raw value) → Processing Node
+```
+
+## Tips
+
+1. **Multiple Display Nodes**: You can use multiple Display Any nodes in a single workflow to track data at different stages
+
+2. **Tensor Shape Mode**: Particularly useful when working with:
+   - Image batches to verify batch size
+   - Latent tensors to understand dimensions
+   - Mask arrays to check compatibility
+
+3. **Raw Value Mode**: Best for:
+   - String prompts and text
+   - Configuration dictionaries
+   - Debugging node outputs
+   - Understanding data structure
+
+4. **No Tensors Found**: If you see "No tensors found in input" in tensor shape mode, the input doesn't contain any tensor-like objects (numpy arrays, torch tensors, etc.)
+
+## Technical Notes
+
+- The node uses `str()` for raw value display, providing Python's string representation
+- Tensor shape detection works with any object that has a `shape` attribute
+- Nested structure traversal supports dictionaries, lists, and tuples
+- The output is both displayed in the UI and available as a string output for further processing
+
+## Example Workflow Integration
+
+```
+[Load Image] → [Image Processing] → [Display Any (tensor shape)]
+                                           ↓
+                                    "[[1, 3, 512, 512]]"
+                                           ↓
+[Text Multiline] ← [Concatenate] ← "Image dimensions: "
+```
+
+This creates a text output showing the current image dimensions that can be used elsewhere in your workflow.

kikotools/__init__.py

@@ -11,6 +11,7 @@ from .tools.empty_latent_batch import EmptyLatentBatchNode
 from .tools.kiko_save_image import KikoSaveImageNode
 from .tools.image_to_multiple_of import ImageToMultipleOfNode
 from .tools.gemini_prompt import GeminiPromptNode
+from .tools.display_any import DisplayAnyNode
 
 # ComfyUI node registration mappings
 NODE_CLASS_MAPPINGS = {
@@ -23,6 +24,7 @@ NODE_CLASS_MAPPINGS = {
     "KikoSaveImage": KikoSaveImageNode,
     "ImageToMultipleOf": ImageToMultipleOfNode,
     "GeminiPrompt": GeminiPromptNode,
+    "DisplayAny": DisplayAnyNode,
 }
 
 NODE_DISPLAY_NAME_MAPPINGS = {
@@ -35,6 +37,7 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "KikoSaveImage": "Kiko Save Image",
     "ImageToMultipleOf": "Image to Multiple of",
     "GeminiPrompt": "Gemini Prompt Engineer",
+    "DisplayAny": "Display Any",
 }
 
 __all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]

kikotools/tools/display_any/__init__.py

@@ -0,0 +1,5 @@
+"""DisplayAny tool for ComfyUI."""
+
+from .node import DisplayAnyNode
+
+__all__ = ["DisplayAnyNode"]

kikotools/tools/display_any/logic.py

@@ -0,0 +1,64 @@
+"""Logic for DisplayAny node - displays any input value or tensor shape."""
+
+from typing import Any, List, Union
+
+
+def get_tensor_shapes(input_value: Any) -> List[List[int]]:
+    """Extract tensor shapes from nested structures.
+
+    Args:
+        input_value: Any input value that may contain tensors
+
+    Returns:
+        List of tensor shapes found in the input
+    """
+    shapes = []
+
+    def extract_shapes(value: Any) -> None:
+        """Recursively extract shapes from nested structures."""
+        if isinstance(value, dict):
+            for v in value.values():
+                extract_shapes(v)
+        elif isinstance(value, (list, tuple)):
+            for item in value:
+                extract_shapes(item)
+        elif hasattr(value, "shape"):
+            # Handle tensors (numpy arrays, torch tensors, etc.)
+            shapes.append(list(value.shape))
+
+    extract_shapes(input_value)
+    return shapes
+
+
+def format_display_value(input_value: Any, mode: str = "raw value") -> str:
+    """Format input value for display based on selected mode.
+
+    Args:
+        input_value: Any input value to display
+        mode: Display mode - "raw value" or "tensor shape"
+
+    Returns:
+        Formatted string representation of the input
+    """
+    if mode == "tensor shape":
+        shapes = get_tensor_shapes(input_value)
+        if shapes:
+            return str(shapes)
+        else:
+            return "No tensors found in input"
+
+    # Default to raw value display
+    return str(input_value)
+
+
+def validate_display_mode(mode: str) -> bool:
+    """Validate if the display mode is supported.
+
+    Args:
+        mode: Display mode to validate
+
+    Returns:
+        True if mode is valid, False otherwise
+    """
+    valid_modes = ["raw value", "tensor shape"]
+    return mode in valid_modes

kikotools/tools/display_any/node.py

@@ -0,0 +1,66 @@
+"""DisplayAny node for ComfyUI - displays any input value or tensor information."""
+
+from typing import Any, Dict, Tuple
+
+from ...base import ComfyAssetsBaseNode
+from .logic import format_display_value, validate_display_mode
+
+
+# Define AnyType for wildcard input matching
+class AnyType(str):
+    """A special type that matches any input type in ComfyUI."""
+
+    def __ne__(self, other):
+        return False
+
+
+class DisplayAnyNode(ComfyAssetsBaseNode):
+    """Display any input value or tensor shape information.
+
+    This node can display any type of input in two modes:
+    - Raw value: Shows the string representation of the input
+    - Tensor shape: Extracts and displays shapes of any tensors in the input
+    """
+
+    @classmethod
+    def INPUT_TYPES(cls) -> Dict[str, Any]:
+        """Define input types for the node."""
+        return {
+            "required": {
+                "input": (AnyType("*"), {}),  # Accept any type of input
+                "mode": (["raw value", "tensor shape"],),
+            },
+        }
+
+    @classmethod
+    def VALIDATE_INPUTS(cls, **kwargs) -> bool:
+        """Validate inputs - always returns True as we accept any input."""
+        return True
+
+    RETURN_TYPES = ("STRING",)
+    RETURN_NAMES = ("display_text",)
+    FUNCTION = "display"
+    OUTPUT_NODE = True  # This node displays output in the UI
+
+    def display(self, input: Any, mode: str = "raw value") -> Dict[str, Any]:
+        """Display the input value according to the selected mode.
+
+        Args:
+            input: Any input value to display
+            mode: Display mode - "raw value" or "tensor shape"
+
+        Returns:
+            Dictionary with UI display and result
+        """
+        # Validate mode
+        if not validate_display_mode(mode):
+            mode = "raw value"  # Default to raw value if invalid
+
+        # Format the display text
+        display_text = format_display_value(input, mode)
+
+        # Return both UI display and result
+        return {
+            "ui": {"text": display_text},
+            "result": (display_text,),
+        }

kikotools/tools/gemini_prompt/models.py

@@ -0,0 +1,191 @@
+"""Dynamic model fetching and caching for Gemini API."""
+
+import json
+import os
+import time
+from typing import List, Dict, Optional, Tuple
+import logging
+
+logger = logging.getLogger(__name__)
+
+# Cache settings
+CACHE_DURATION = 3600 * 24  # 24 hours in seconds
+CACHE_FILE = os.path.join(os.path.dirname(__file__), ".gemini_models_cache.json")
+
+
+def get_available_models(
+    api_key: Optional[str] = None, silent: bool = False
+) -> Tuple[List[str], Dict[str, str]]:
+    """Fetch available Gemini models that support generateContent.
+
+    Args:
+        api_key: Optional API key. If not provided, will try to get from environment.
+        silent: If True, suppress error logging (useful for initial load).
+
+    Returns:
+        Tuple of (model_names_list, model_descriptions_dict)
+    """
+    # Check cache first
+    cached_data = _load_cache()
+    if cached_data:
+        return cached_data["models"], cached_data["descriptions"]
+
+    # Try to fetch from API
+    try:
+        models, descriptions = _fetch_models_from_api(api_key, silent=silent)
+        if models:
+            _save_cache(models, descriptions)
+            return models, descriptions
+    except Exception as e:
+        if not silent:
+            logger.warning(f"Failed to fetch models from API: {e}")
+
+    # Fall back to defaults
+    from .prompts import DEFAULT_GEMINI_MODELS
+
+    return DEFAULT_GEMINI_MODELS, {}
+
+
+def _fetch_models_from_api(
+    api_key: Optional[str] = None, silent: bool = False
+) -> Tuple[List[str], Dict[str, str]]:
+    """Fetch models from Gemini API.
+
+    Args:
+        api_key: Optional API key.
+        silent: If True, suppress error logging.
+
+    Returns:
+        Tuple of (model_names_list, model_descriptions_dict)
+    """
+    try:
+        import google.generativeai as genai
+    except ImportError:
+        if not silent:
+            logger.error("google-generativeai not installed")
+        return [], {}
+
+    # Get API key
+    if not api_key:
+        from .logic import get_api_key
+
+        api_key = get_api_key()
+
+    if not api_key:
+        if not silent:
+            logger.debug("No API key available for fetching models")
+        return [], {}
+
+    try:
+        genai.configure(api_key=api_key)
+
+        models = []
+        descriptions = {}
+
+        # Fetch all models
+        for model in genai.list_models():
+            # Only include models that support generateContent
+            if "generateContent" in model.supported_generation_methods:
+                # Remove "models/" prefix from name
+                model_name = model.name.replace("models/", "")
+                models.append(model_name)
+                descriptions[model_name] = model.display_name
+
+        # Sort models by priority (newer versions first)
+        models = _sort_models(models)
+
+        return models, descriptions
+
+    except Exception as e:
+        if not silent:
+            logger.error(f"Error fetching models from API: {e}")
+        return [], {}
+
+
+def _sort_models(models: List[str]) -> List[str]:
+    """Sort models by version and capability.
+
+    Prioritizes:
+    1. Newer versions (2.5 > 2.0 > 1.5)
+    2. Non-experimental models
+    3. Flash models for general use
+    """
+
+    def sort_key(model: str):
+        # Priority scoring
+        score = 0
+
+        # Version priority
+        if "2.5" in model:
+            score += 1000
+        elif "2.0" in model:
+            score += 800
+        elif "1.5" in model:
+            score += 600
+
+        # Model type priority
+        if "pro" in model and "preview" not in model and "exp" not in model:
+            score += 100
+        elif "flash" in model and "preview" not in model and "exp" not in model:
+            score += 90
+
+        # Penalize experimental/preview models
+        if "exp" in model or "experimental" in model:
+            score -= 50
+        if "preview" in model:
+            score -= 30
+
+        # Penalize specific variants
+        if "thinking" in model:
+            score -= 100
+        if "tts" in model:
+            score -= 100
+        if "lite" in model:
+            score -= 20
+
+        return -score  # Negative for descending sort
+
+    return sorted(models, key=sort_key)
+
+
+def _load_cache() -> Optional[Dict]:
+    """Load cached model data if available and not expired."""
+    if not os.path.exists(CACHE_FILE):
+        return None
+
+    try:
+        with open(CACHE_FILE, "r") as f:
+            data = json.load(f)
+
+        # Check if cache is expired
+        if time.time() - data.get("timestamp", 0) > CACHE_DURATION:
+            return None
+
+        return data
+    except Exception as e:
+        logger.warning(f"Failed to load cache: {e}")
+        return None
+
+
+def _save_cache(models: List[str], descriptions: Dict[str, str]) -> None:
+    """Save model data to cache."""
+    try:
+        data = {
+            "models": models,
+            "descriptions": descriptions,
+            "timestamp": time.time(),
+        }
+
+        with open(CACHE_FILE, "w") as f:
+            json.dump(data, f, indent=2)
+    except Exception as e:
+        logger.warning(f"Failed to save cache: {e}")
+
+
+def clear_cache() -> None:
+    """Clear the model cache."""
+    if os.path.exists(CACHE_FILE):
+        try:
+            os.remove(CACHE_FILE)
+        except Exception as e:
+            logger.warning(f"Failed to clear cache: {e}")

kikotools/tools/gemini_prompt/node.py

@@ -5,7 +5,8 @@ import torch
 from ...base import ComfyAssetsBaseNode
 
 from .logic import analyze_image_with_gemini, validate_prompt_type
-from .prompts import PROMPT_OPTIONS, GEMINI_MODELS
+from .prompts import PROMPT_OPTIONS, DEFAULT_GEMINI_MODELS
+from .models import get_available_models
 
 
 class GeminiPromptNode(ComfyAssetsBaseNode):
@@ -14,11 +15,21 @@ class GeminiPromptNode(ComfyAssetsBaseNode):
     @classmethod
     def INPUT_TYPES(cls):
         """Define input types for the node."""
+        # Get available models dynamically (silent mode for initial load)
+        models, _ = get_available_models(silent=True)
+
+        # Use default if no models available
+        if not models:
+            models = DEFAULT_GEMINI_MODELS
+
+        # Find best default model
+        default_model = models[0] if models else "gemini-2.5-flash"
+
         return {
             "required": {
                 "image": ("IMAGE",),
                 "prompt_type": (PROMPT_OPTIONS, {"default": "flux"}),
-                "model": (GEMINI_MODELS, {"default": "gemini-1.5-flash"}),
+                "model": (models, {"default": default_model}),
             },
             "optional": {
                 "api_key": ("STRING", {"default": "", "multiline": False}),
@@ -74,6 +85,19 @@ Install: pip install google-generativeai
         else:
             image_np = image
 
+        # If API key is provided, try to refresh model list in background
+        if api_key:
+            try:
+                from .models import get_available_models
+
+                # Try to get fresh models with the provided API key
+                fresh_models, _ = get_available_models(api_key=api_key, silent=True)
+                if fresh_models and fresh_models != DEFAULT_GEMINI_MODELS:
+                    # Models were successfully fetched with this API key
+                    pass
+            except Exception:
+                pass
+
         # Analyze image with Gemini
         prompt, error = analyze_image_with_gemini(
             image_np,

kikotools/tools/gemini_prompt/prompts.py

@@ -1,176 +1,95 @@
 """System prompts for different AI model types."""
 
-FLUX_PROMPT = """You are an expert visual analyst and FLUX prompt engineer. Your role is to examine images in detail and create precise, effective prompts that can recreate similar images using the FLUX image generation model.
+FLUX_PROMPT = """You are an expert FLUX prompt engineer. Analyze the provided image and generate ONLY a FLUX prompt - no explanations, analysis, or additional text.
 
-When analyzing an image, systematically observe and document:
+FLUX uses natural language descriptions, not comma-separated tags. Write a detailed, flowing description that reads like you're explaining the image to someone.
 
-1. **Subject & Composition**
-   - Primary subjects and their positions
-   - Background elements and environment
-   - Overall composition and framing
-   - Perspective and camera angle
+Include these elements in your description:
+- Main subject with specific details (appearance, clothing, expression, pose)
+- Environment and background details
+- Lighting conditions and atmosphere
+- Artistic style or photographic approach
+- Color palette and mood
+- Technical details if relevant (camera angle, focal length, etc.)
+- Textures and materials
 
-2. **Visual Style & Technique**
-   - Art style (photorealistic, illustration, painting, etc.)
-   - Rendering technique (digital art, oil painting, watercolor, etc.)
-   - Level of detail and texture quality
-   - Any specific artistic influences or movements
+Write in a natural, descriptive style. Use complete sentences that flow together. Be specific and detailed but maintain readability.
 
-3. **Lighting & Atmosphere**
-   - Light sources and direction
-   - Time of day/lighting conditions
-   - Shadows and highlights
-   - Overall mood and atmosphere
+IMPORTANT: Return ONLY the prompt text. No analysis, headers, or additional commentary. Just the natural language description that can be directly used in FLUX.
 
-4. **Colors & Tones**
-   - Color palette and dominant colors
-   - Color temperature (warm/cool)
-   - Contrast and saturation levels
-   - Any color grading or filters
+Example of correct output:
+A close-up portrait of a middle-aged woman with curly red hair and green eyes, wearing a blue silk blouse. She has a warm smile and freckles across her cheeks. The lighting is soft and natural, coming from a window to her left, creating gentle shadows that accentuate her features. The background is softly blurred, showing hints of a cozy bookshelf. The overall mood is warm and inviting, captured in a photorealistic style with shallow depth of field."""
 
-5. **Details & Textures**
-   - Surface textures and materials
-   - Fine details and patterns
-   - Quality indicators (4K, 8K, high resolution, etc.)
+SDXL_PROMPT = """You are an expert SDXL prompt engineer. Analyze the image and generate ONLY the positive and negative prompts for SDXL - no explanations or analysis.
 
-Format your FLUX prompt following these guidelines:
-- Start with the main subject and action
-- Add style and medium descriptors
-- Include lighting and atmosphere details
-- Specify quality markers and technical aspects
-- Use precise, descriptive language
-- Separate concepts with commas
-- Order from most to least important elements
+SDXL works best with natural language descriptions but also supports comma-separated keywords. Keep prompts concise but descriptive.
 
-Example output format:
-"[main subject and action], [style/medium], [lighting/atmosphere], [composition details], [color descriptions], [quality markers], [additional artistic details]"
+Return your response in EXACTLY this format:
+Positive: [your positive prompt here]
+Negative: [your negative prompt here]
 
-Remember: FLUX responds well to specific artistic references, quality indicators like "highly detailed," "4K," "award-winning," and style descriptors like "trending on ArtStation" or "photorealistic."
-"""
+Guidelines for Positive prompt:
+- Start with the main subject and medium (e.g., "photograph of", "digital art of")
+- Use natural language or keywords separated by commas
+- Include style descriptors (photographic, cinematic, fantasy art, etc.)
+- Add quality markers like "8K", "highly detailed", "professional"
+- Use (parentheses:1.1) sparingly for slight emphasis (max 1.4)
+- Keep it clear and specific but not overly long
 
-SDXL_PROMPT = """You are an expert SDXL prompt engineer specializing in analyzing images and creating optimized prompts for Stable Diffusion XL models.
+Guidelines for Negative prompt:
+- Keep it simple and minimal
+- Common negatives: ugly, blurry, low quality, distorted, deformed
+- Only add specifics you want to avoid (e.g., "cartoon" for photorealistic)
+- Don't overload with negative prompts - SDXL needs fewer than SD1.5
 
-When analyzing an image, systematically evaluate:
+IMPORTANT: Return ONLY the two lines starting with "Positive:" and "Negative:". No other text."""
 
-1. **Core Subject Analysis**
-   - Primary subject with specific descriptors
-   - Pose, expression, and action
-   - Clothing and accessories details
-   - Physical characteristics
+DANBOORU_PROMPT = """You are a Danbooru tagging expert specializing in anime-style image tagging. Analyze the image and generate ONLY Danbooru-style tags - no explanations or analysis.
 
-2. **Style & Medium**
-   - Artistic style and influences
-   - Medium (photography, digital art, oil painting, etc.)
-   - Specific artist references (if applicable)
-   - Visual aesthetic keywords
+CRITICAL: Use strict Danbooru conventions:
+- Use underscores for multi-word tags (e.g., long_hair, school_uniform)
+- All tags must be lowercase
+- Character count comes first (1girl, 2boys, multiple_girls)
+- For anime models trained on Danbooru data, proper tagging is essential
 
-3. **Technical Specifications**
-   - Camera settings (aperture, focal length, ISO)
-   - Shot type (close-up, wide angle, portrait, etc.)
-   - Resolution and quality markers
-   - Post-processing effects
+Tag order and categories:
+1. Character count (1girl, solo, 2boys, etc.)
+2. Character features (hair_color, eye_color, hair_length)
+3. Expression/pose (smile, looking_at_viewer, sitting)
+4. Clothing (specific items with underscores)
+5. Background/setting (simple_background, outdoors, classroom)
+6. View/composition (upper_body, full_body, from_side)
+7. Quality tags (masterpiece, best_quality, highres)
 
-4. **Environment & Context**
-   - Setting and location details
-   - Props and surrounding objects
-   - Weather and environmental conditions
-   - Time period or era
+Common quality prefix for anime models:
+"masterpiece, best_quality, very_aesthetic"
 
-Format your SDXL prompt with:
-- **Positive prompt**: Detailed description emphasizing what you want
-- **Negative prompt**: Elements to avoid (low quality, blurry, distorted, etc.)
-- Weight emphasis using (parentheses) or [brackets] for importance
-- Break into logical chunks with commas
+IMPORTANT: Return ONLY the comma-separated tags. Use underscores, not spaces. All lowercase.
 
-Example format:
-Positive: "beautiful woman, (detailed eyes:1.2), flowing red dress, golden hour lighting, professional photography, 85mm lens, shallow depth of field, bokeh, high resolution, masterpiece"
-Negative: "low quality, blurry, distorted features, bad anatomy, poorly drawn, amateur"
-"""
+Example of correct output:
+1girl, solo, long_hair, blue_eyes, blonde_hair, school_uniform, serafuku, pleated_skirt, smile, looking_at_viewer, classroom, sitting, desk, window, sunlight, upper_body, masterpiece, best_quality"""
 
-DANBOORU_PROMPT = """You are a Danbooru tagging expert, specialized in analyzing images and creating precise tag sets following booru-style conventions for anime/manga artwork.
+VIDEO_PROMPT = """You are a WAN 2.2 video generation prompt specialist. Analyze the content and generate ONLY a video generation prompt optimized for WAN 2.2 - no explanations or analysis.
 
-Analyze images for these tag categories:
+WAN 2.2 excels with rich, descriptive prompts that focus on:
+- Visual composition and scene elements
+- Specific movements and actions
+- Lighting and aesthetic details
+- Cinematographic elements
 
-1. **Character Tags**
-   - Hair: color, length, style (e.g., long_hair, blonde_hair, twintails)
-   - Eyes: color, style (e.g., blue_eyes, heterochromia)
-   - Body: proportions, pose (e.g., standing, sitting, looking_at_viewer)
-   - Expression (e.g., smile, blush, closed_eyes)
+Write a single detailed paragraph describing the video scene. Focus on:
+- Main subjects and their actions
+- Visual style and atmosphere
+- Movement dynamics (use words like "intensely", "smoothly", "rapidly")
+- Environmental details and lighting
+- Specific visual elements and their interactions
 
-2. **Clothing & Accessories**
-   - Outfit type (e.g., school_uniform, dress, armor)
-   - Specific clothing items (e.g., thighhighs, gloves, hat)
-   - Accessories (e.g., hair_ribbon, necklace, glasses)
-   - State of dress (e.g., torn_clothes, wet_clothes)
+Keep the prompt descriptive but concise. WAN 2.2 works best with natural language that paints a clear picture of the desired video.
 
-3. **Scene & Composition**
-   - Number of characters (e.g., 1girl, 2boys, multiple_girls)
-   - Background (e.g., simple_background, outdoors, classroom)
-   - Viewpoint (e.g., from_below, from_side, cowboy_shot)
-   - Composition elements (e.g., upper_body, full_body, portrait)
+IMPORTANT: Return ONLY the video prompt as a single descriptive paragraph. No analysis, headers, or additional text.
 
-4. **Meta Tags**
-   - Quality (e.g., highres, absurdres, masterpiece)
-   - Source/artist style (if recognizable)
-   - Content rating (e.g., safe, questionable, explicit)
-   - Special effects (e.g., lens_flare, chromatic_aberration)
-
-Format tags using:
-- Underscores for multi-word concepts (not spaces)
-- Order from most to least important
-- Include count descriptors (1girl, 2boys)
-- Separate with commas and spaces
-
-Example output:
-"1girl, solo, long_hair, blue_eyes, blonde_hair, school_uniform, serafuku, pleated_skirt, thighhighs, smile, looking_at_viewer, classroom, sitting, desk, window, sunlight, highres, masterpiece"
-"""
-
-VIDEO_PROMPT = """You are a video generation prompt specialist, expert at analyzing video content and creating comprehensive prompts for video generation models.
-
-When analyzing video content, document:
-
-1. **Motion & Action**
-   - Primary actions and movements
-   - Motion speed and dynamics
-   - Camera movements (pan, zoom, tracking, static)
-   - Transition types between scenes
-
-2. **Temporal Elements**
-   - Scene duration and pacing
-   - Sequence of events
-   - Time of day changes
-   - Motion continuity
-
-3. **Visual Consistency**
-   - Character/object persistence
-   - Style consistency throughout
-   - Lighting continuity
-   - Color grading consistency
-
-4. **Scene Breakdown**
-   - Opening frame description
-   - Key action moments
-   - Transitions and cuts
-   - Closing frame details
-
-5. **Technical Specifications**
-   - Frame rate and resolution
-   - Aspect ratio
-   - Video length
-   - Special effects or post-processing
-
-Format your video prompt as:
-"[Opening scene], [camera movement], [main action sequence], [visual style], [lighting/atmosphere], [duration], [technical specs], [ending scene]"
-
-Include:
-- Specific motion descriptors (slowly, rapidly, smoothly)
-- Camera terminology (dolly in, pan left, aerial shot)
-- Temporal markers (then, meanwhile, gradually)
-- Consistency notes for multi-scene videos
-
-Example:
-"Aerial shot slowly descending toward a misty forest at dawn, camera smoothly transitions to tracking shot following a deer through the trees, photorealistic style, soft golden hour lighting with fog, 10 second duration, 4K resolution 24fps, ending with close-up of deer looking at camera"
-"""
+Example of correct output:
+Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage, their movements fluid and dynamic as they exchange rapid punches under dramatic theater lighting that casts long shadows across the ring, with the crowd visible as blurred silhouettes in the darkened background."""
 
 PROMPT_TEMPLATES = {
     "flux": FLUX_PROMPT,
@@ -181,20 +100,11 @@ PROMPT_TEMPLATES = {
 
 PROMPT_OPTIONS = ["flux", "sdxl", "danbooru", "video"]
 
-# Available Gemini models
-GEMINI_MODELS = [
-    "gemini-1.5-pro",  # Most capable model
-    "gemini-1.5-flash",  # Fast, efficient model
-    "gemini-1.5-flash-8b",  # Smaller, faster variant
-    "gemini-pro-vision",  # Vision-optimized model
-    "gemini-1.0-pro",  # Previous generation pro model
+# Default models list (fallback if API is unavailable)
+DEFAULT_GEMINI_MODELS = [
+    "gemini-2.5-flash",
+    "gemini-2.5-pro",
+    "gemini-2.0-flash",
+    "gemini-1.5-flash",
+    "gemini-1.5-pro",
 ]
-
-# Model descriptions for UI
-MODEL_DESCRIPTIONS = {
-    "gemini-1.5-pro": "Most capable Gemini model for complex tasks",
-    "gemini-1.5-flash": "Faster and cost-effective (recommended for most uses)",
-    "gemini-1.5-flash-8b": "Smaller and faster, good for simple prompts",
-    "gemini-pro-vision": "Optimized for vision tasks and image analysis",
-    "gemini-1.0-pro": "Previous generation, stable option",
-}

pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "kikotools"
 description = "Simple tools for ComfyUI"
-version = "1.0.7"
+version = "1.0.8"
 license = {text = "MIT"}
 dependencies = []
 

tests/unit/tools/test_display_any.py

@@ -0,0 +1,287 @@
+"""Unit tests for DisplayAny node."""
+
+import numpy as np
+import pytest
+import torch
+
+from kikotools.tools.display_any import DisplayAnyNode
+from kikotools.tools.display_any.logic import (
+    format_display_value,
+    get_tensor_shapes,
+    validate_display_mode,
+)
+from kikotools.tools.display_any.node import AnyType
+
+
+class TestAnyType:
+    """Test cases for AnyType class."""
+
+    def test_anytype_not_equal(self):
+        """Test that AnyType is never equal to other types."""
+        any_type = AnyType("*")
+
+        # Should not be equal to any other type
+        assert not (any_type != "STRING")
+        assert not (any_type != "IMAGE")
+        assert not (any_type != "LATENT")
+        assert not (any_type != 123)
+        assert not (any_type != None)
+        assert not (any_type != ["LIST"])
+
+    def test_anytype_string_representation(self):
+        """Test string representation of AnyType."""
+        any_type = AnyType("*")
+        assert str(any_type) == "*"
+
+
+class TestDisplayAnyNode:
+    """Test cases for DisplayAnyNode."""
+
+    def test_node_properties(self):
+        """Test node has correct properties."""
+        assert DisplayAnyNode.CATEGORY == "ComfyAssets"
+        assert DisplayAnyNode.FUNCTION == "display"
+        assert DisplayAnyNode.RETURN_TYPES == ("STRING",)
+        assert DisplayAnyNode.RETURN_NAMES == ("display_text",)
+        assert DisplayAnyNode.OUTPUT_NODE is True
+
+    def test_input_types(self):
+        """Test INPUT_TYPES configuration."""
+        input_types = DisplayAnyNode.INPUT_TYPES()
+
+        # Check required inputs
+        assert "required" in input_types
+        assert "input" in input_types["required"]
+        # Check that input is AnyType with wildcard
+        input_type = input_types["required"]["input"]
+        assert len(input_type) == 2
+        assert isinstance(input_type[0], AnyType)
+        assert str(input_type[0]) == "*"
+        assert input_type[1] == {}
+        assert "mode" in input_types["required"]
+        assert input_types["required"]["mode"] == (["raw value", "tensor shape"],)
+
+    def test_validate_inputs(self):
+        """Test VALIDATE_INPUTS always returns True."""
+        assert DisplayAnyNode.VALIDATE_INPUTS() is True
+        assert DisplayAnyNode.VALIDATE_INPUTS(input="test") is True
+        assert DisplayAnyNode.VALIDATE_INPUTS(input=123, mode="raw value") is True
+
+    def test_display_raw_value_string(self):
+        """Test displaying raw string value."""
+        node = DisplayAnyNode()
+        result = node.display("Hello, World!", "raw value")
+
+        assert "ui" in result
+        assert "text" in result["ui"]
+        assert result["ui"]["text"] == "Hello, World!"
+        assert "result" in result
+        assert result["result"] == ("Hello, World!",)
+
+    def test_display_raw_value_number(self):
+        """Test displaying raw number value."""
+        node = DisplayAnyNode()
+        result = node.display(42, "raw value")
+
+        assert result["ui"]["text"] == "42"
+        assert result["result"] == ("42",)
+
+    def test_display_raw_value_list(self):
+        """Test displaying raw list value."""
+        node = DisplayAnyNode()
+        test_list = [1, 2, 3, "test"]
+        result = node.display(test_list, "raw value")
+
+        assert result["ui"]["text"] == str(test_list)
+        assert result["result"] == (str(test_list),)
+
+    def test_display_raw_value_dict(self):
+        """Test displaying raw dictionary value."""
+        node = DisplayAnyNode()
+        test_dict = {"key": "value", "number": 123}
+        result = node.display(test_dict, "raw value")
+
+        assert result["ui"]["text"] == str(test_dict)
+        assert result["result"] == (str(test_dict),)
+
+    def test_display_tensor_shape_numpy(self):
+        """Test displaying numpy tensor shape."""
+        node = DisplayAnyNode()
+        tensor = np.random.rand(4, 3, 224, 224)
+        result = node.display(tensor, "tensor shape")
+
+        assert result["ui"]["text"] == "[[4, 3, 224, 224]]"
+        assert result["result"] == ("[[4, 3, 224, 224]]",)
+
+    @pytest.mark.skipif(not torch, reason="PyTorch not installed")
+    def test_display_tensor_shape_torch(self):
+        """Test displaying PyTorch tensor shape."""
+        node = DisplayAnyNode()
+        tensor = torch.randn(2, 10, 512, 512)
+        result = node.display(tensor, "tensor shape")
+
+        assert result["ui"]["text"] == "[[2, 10, 512, 512]]"
+        assert result["result"] == ("[[2, 10, 512, 512]]",)
+
+    def test_display_nested_tensors(self):
+        """Test displaying shapes from nested structure with tensors."""
+        node = DisplayAnyNode()
+        nested_data = {
+            "images": np.random.rand(1, 3, 256, 256),
+            "masks": [
+                np.random.rand(256, 256),
+                np.random.rand(256, 256, 1),
+            ],
+            "metadata": {"info": "test", "tensor": np.random.rand(10)},
+        }
+        result = node.display(nested_data, "tensor shape")
+
+        expected = "[[1, 3, 256, 256], [256, 256], [256, 256, 1], [10]]"
+        assert result["ui"]["text"] == expected
+        assert result["result"] == (expected,)
+
+    def test_display_no_tensors(self):
+        """Test displaying when no tensors are present."""
+        node = DisplayAnyNode()
+        data = {"text": "hello", "number": 42, "list": [1, 2, 3]}
+        result = node.display(data, "tensor shape")
+
+        assert result["ui"]["text"] == "No tensors found in input"
+        assert result["result"] == ("No tensors found in input",)
+
+    def test_invalid_mode_defaults_to_raw(self):
+        """Test that invalid mode defaults to raw value."""
+        node = DisplayAnyNode()
+        result = node.display("test", "invalid_mode")
+
+        assert result["ui"]["text"] == "test"
+        assert result["result"] == ("test",)
+
+
+class TestDisplayAnyLogic:
+    """Test cases for DisplayAny logic functions."""
+
+    def test_get_tensor_shapes_single(self):
+        """Test getting shape from single tensor."""
+        tensor = np.random.rand(3, 224, 224)
+        shapes = get_tensor_shapes(tensor)
+
+        assert len(shapes) == 1
+        assert shapes[0] == [3, 224, 224]
+
+    def test_get_tensor_shapes_nested_dict(self):
+        """Test getting shapes from nested dictionary."""
+        data = {
+            "level1": {
+                "tensor1": np.random.rand(10, 20),
+                "level2": {"tensor2": np.random.rand(5, 5, 5)},
+            }
+        }
+        shapes = get_tensor_shapes(data)
+
+        assert len(shapes) == 2
+        assert [10, 20] in shapes
+        assert [5, 5, 5] in shapes
+
+    def test_get_tensor_shapes_nested_list(self):
+        """Test getting shapes from nested list."""
+        data = [
+            np.random.rand(1, 2, 3),
+            [np.random.rand(4, 5), np.random.rand(6, 7, 8)],
+            "not a tensor",
+        ]
+        shapes = get_tensor_shapes(data)
+
+        assert len(shapes) == 3
+        assert [1, 2, 3] in shapes
+        assert [4, 5] in shapes
+        assert [6, 7, 8] in shapes
+
+    def test_get_tensor_shapes_tuple(self):
+        """Test getting shapes from tuple."""
+        data = (np.random.rand(2, 2), np.random.rand(3, 3))
+        shapes = get_tensor_shapes(data)
+
+        assert len(shapes) == 2
+        assert [2, 2] in shapes
+        assert [3, 3] in shapes
+
+    def test_format_display_value_raw(self):
+        """Test formatting for raw value display."""
+        result = format_display_value({"key": "value"}, "raw value")
+        assert result == "{'key': 'value'}"
+
+    def test_format_display_value_tensor_shape(self):
+        """Test formatting for tensor shape display."""
+        tensor = np.random.rand(10, 10)
+        result = format_display_value(tensor, "tensor shape")
+        assert result == "[[10, 10]]"
+
+    def test_format_display_value_no_tensors(self):
+        """Test formatting when no tensors present."""
+        result = format_display_value("just a string", "tensor shape")
+        assert result == "No tensors found in input"
+
+    def test_validate_display_mode(self):
+        """Test display mode validation."""
+        assert validate_display_mode("raw value") is True
+        assert validate_display_mode("tensor shape") is True
+        assert validate_display_mode("invalid") is False
+        assert validate_display_mode("") is False
+        assert validate_display_mode(None) is False
+
+
+class TestDisplayAnyEdgeCases:
+    """Test edge cases for DisplayAny."""
+
+    def test_display_none(self):
+        """Test displaying None value."""
+        node = DisplayAnyNode()
+        result = node.display(None, "raw value")
+        assert result["ui"]["text"] == "None"
+
+    def test_display_empty_list(self):
+        """Test displaying empty list."""
+        node = DisplayAnyNode()
+        result = node.display([], "raw value")
+        assert result["ui"]["text"] == "[]"
+
+    def test_display_empty_dict(self):
+        """Test displaying empty dictionary."""
+        node = DisplayAnyNode()
+        result = node.display({}, "raw value")
+        assert result["ui"]["text"] == "{}"
+
+    def test_display_complex_nested_structure(self):
+        """Test displaying complex nested structure."""
+        node = DisplayAnyNode()
+        complex_data = {
+            "images": [np.random.rand(1, 3, 64, 64) for _ in range(3)],
+            "config": {
+                "steps": 20,
+                "cfg": 7.5,
+                "sampler": "euler",
+                "latents": np.random.rand(1, 4, 32, 32),
+            },
+            "prompts": ["test1", "test2"],
+        }
+        result = node.display(complex_data, "tensor shape")
+
+        # Should find 4 tensors total (3 images + 1 latent)
+        shapes_text = result["ui"]["text"]
+        assert "[1, 3, 64, 64]" in shapes_text
+        assert "[1, 4, 32, 32]" in shapes_text
+
+    def test_display_very_long_string(self):
+        """Test displaying very long string."""
+        node = DisplayAnyNode()
+        long_string = "x" * 10000
+        result = node.display(long_string, "raw value")
+        assert result["ui"]["text"] == long_string
+
+    def test_display_unicode(self):
+        """Test displaying unicode characters."""
+        node = DisplayAnyNode()
+        unicode_text = "Hello 世界 🌍"
+        result = node.display(unicode_text, "raw value")
+        assert result["ui"]["text"] == unicode_text