@nockdev/awf 6.2.0 → 6.2.1

package/.agent/skills/ai-ml/gemini-media-gen/data/files-api.yaml

@@ -0,0 +1,96 @@
+# =============================================================================
+# files-api.yaml — Gemini Files API Patterns
+# =============================================================================
+# 6 patterns for file upload, management, and multimodal usage
+# Source: https://ai.google.dev/gemini-api/docs/files
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 6
+
+overview:
+  storage: "20GB per project"
+  max_file_size: "2GB"
+  retention: "48 hours (auto-deleted)"
+  cost: "Free"
+  download: "Cannot download uploaded files"
+
+supported_types:
+  images: ["image/jpeg", "image/png", "image/gif", "image/webp", "image/heic"]
+  video:
+    [
+      "video/mp4",
+      "video/mpeg",
+      "video/quicktime",
+      "video/webm",
+      "video/x-msvideo",
+    ]
+  audio:
+    [
+      "audio/aac",
+      "audio/flac",
+      "audio/mp3",
+      "audio/m4a",
+      "audio/ogg",
+      "audio/wav",
+    ]
+  documents: ["application/pdf", "text/plain", "text/csv", "text/html"]
+
+operations:
+  upload:
+    desc: "Upload a file for use in generation"
+    python: |
+      from google import genai
+      client = genai.Client(api_key="API_KEY")
+      file = client.files.upload(file="video.mp4")
+      print(f"URI: {file.uri}, State: {file.state}")
+    javascript: |
+      const file = await ai.files.upload({ file: "video.mp4" });
+      console.log(`URI: ${file.uri}, State: ${file.state}`);
+
+  check_status:
+    desc: "Wait for video/large file processing"
+    python: |
+      import time
+      while file.state.name == "PROCESSING":
+          time.sleep(5)
+          file = client.files.get(name=file.name)
+      if file.state.name == "FAILED":
+          raise ValueError("File processing failed")
+
+  use_in_generation:
+    desc: "Reference uploaded file in model request"
+    python: |
+      response = client.models.generate_content(
+          model="gemini-2.5-flash",
+          contents=[file, "Summarize this video in 3 bullet points"]
+      )
+
+  list_files:
+    desc: "List all uploaded files"
+    python: |
+      for f in client.files.list():
+          print(f"{f.name}: {f.size_bytes} bytes, {f.state}")
+
+  get_metadata:
+    desc: "Get file metadata by name"
+    python: |
+      file = client.files.get(name="files/abc123")
+      print(f"Name: {file.display_name}, Types: {file.mime_type}")
+
+  delete_file:
+    desc: "Delete a file before 48h auto-expiry"
+    python: |
+      client.files.delete(name=file.name)
+
+best_practices:
+  - "Upload large files (>20MB) via Files API instead of inline"
+  - "Video processing takes 30s-5min depending on size"
+  - "Check state before using: PROCESSING → ACTIVE → use in request"
+  - "Delete sensitive files immediately after use"
+  - "Use display_name for easier file management"
+  - "Files API works across all Gemini models"
+
+# =============================================================================
+# _DOMYH Awesome Code • Files API v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-media-gen/data/image-models.yaml

@@ -0,0 +1,112 @@
+# =============================================================================
+# image-models.yaml — Nano Banana Image Generation Models
+# =============================================================================
+# 8 patterns for model selection, configuration, and API usage
+# Source: https://ai.google.dev/gemini-api/docs/image-generation
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 8
+
+models:
+  flash:
+    id: "gemini-2.0-flash-preview-image-generation"
+    display: "Nano Banana Flash"
+    tier: "speed"
+    max_resolution: "1024x1024"
+    max_reference_images: 5
+    thinking: false
+    search_grounding: false
+    text_rendering: "basic"
+    synthid: true
+    best_for:
+      - "Rapid prototyping"
+      - "Batch generation"
+      - "Low-latency applications"
+      - "Standard quality images"
+    config:
+      response_modalities: ["TEXT", "IMAGE"]
+      temperature: 1.0 # Recommended for image gen
+      top_p: null # Not supported for image gen
+      top_k: null # Not supported for image gen
+
+  pro:
+    id: "gemini-2.5-pro-preview-image-generation"
+    display: "Nano Banana Pro"
+    tier: "quality"
+    max_resolution: "4096x4096"
+    max_reference_images: 14
+    thinking: true
+    search_grounding: true
+    text_rendering: "enhanced"
+    synthid: true
+    best_for:
+      - "Production-quality images"
+      - "4K resolution output"
+      - "Complex multi-reference compositing"
+      - "Accurate text rendering"
+      - "Search-grounded image generation"
+    config:
+      response_modalities: ["TEXT", "IMAGE"]
+      temperature: 1.0
+      thinking_config:
+        thinking_budget: 4096 # Optional, for complex prompts
+
+aspect_ratios:
+  - ratio: "1:1"
+    desc: "Square (default)"
+    use_case: "Social media, profile pictures, icons"
+  - ratio: "16:9"
+    desc: "Wide landscape"
+    use_case: "Desktop wallpapers, YouTube thumbnails, presentations"
+  - ratio: "9:16"
+    desc: "Tall portrait"
+    use_case: "Mobile wallpapers, Instagram stories, TikTok covers"
+  - ratio: "4:3"
+    desc: "Standard landscape"
+    use_case: "Photo prints, blog images"
+  - ratio: "3:4"
+    desc: "Standard portrait"
+    use_case: "Product photos, book covers"
+
+safety_settings:
+  description: "All generated images include invisible SynthID watermark"
+  blocked_content:
+    - "CSAM (Child Sexual Abuse Material)"
+    - "Violent or graphic content"
+    - "Hate speech imagery"
+  person_generation:
+    default: "ALLOW_ADULT"
+    options:
+      - "DONT_ALLOW"
+      - "ALLOW_ADULT"
+
+api_usage:
+  python: |
+    from google import genai
+    client = genai.Client(api_key="YOUR_API_KEY")
+    response = client.models.generate_content(
+        model="gemini-2.0-flash-preview-image-generation",
+        contents="A serene Japanese garden with cherry blossoms",
+        config=genai.types.GenerateContentConfig(
+            response_modalities=["TEXT", "IMAGE"]
+        )
+    )
+    # Access image: response.candidates[0].content.parts[0].inline_data
+  javascript: |
+    const response = await ai.models.generateContent({
+        model: "gemini-2.0-flash-preview-image-generation",
+        contents: "A serene Japanese garden with cherry blossoms",
+        config: { responseModalities: ["TEXT", "IMAGE"] }
+    });
+
+batch_generation:
+  description: "Generate multiple images in sequence"
+  tips:
+    - "Use chat mode to maintain style consistency"
+    - "Reference previous images: 'Generate another image like the previous one but...'"
+    - "Flash model for batch, Pro for final selection"
+
+# =============================================================================
+# _DOMYH Awesome Code • Image Models v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-media-gen/data/image-prompts.yaml

@@ -0,0 +1,131 @@
+# =============================================================================
+# image-prompts.yaml — Image Generation Prompt Patterns
+# =============================================================================
+# 12 patterns for crafting effective image generation prompts
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 12
+
+golden_rule:
+  principle: "Describe the SCENE, don't list KEYWORDS"
+  good: "A golden retriever playing fetch on a sandy beach at sunset, waves crashing in the background"
+  bad: "golden retriever, beach, sunset, waves, fetch, sand"
+
+prompt_structure:
+  template: "[Subject] [Action] [Setting] [Style] [Lighting] [Details]"
+  elements:
+    - name: "Subject"
+      desc: "Who or what is the main focus"
+      example: "A young woman in a red dress"
+    - name: "Action"
+      desc: "What is happening"
+      example: "walking through a field of sunflowers"
+    - name: "Setting"
+      desc: "Where the scene takes place"
+      example: "in a vast countryside under open sky"
+    - name: "Style"
+      desc: "Art direction and aesthetic"
+      examples:
+        - "photorealistic"
+        - "oil painting"
+        - "watercolor"
+        - "digital art"
+        - "anime"
+        - "pencil sketch"
+        - "3D render"
+        - "isometric"
+        - "flat design"
+        - "pixel art"
+    - name: "Lighting"
+      desc: "Light quality and direction"
+      examples:
+        - "golden hour sunlight"
+        - "soft diffused studio light"
+        - "dramatic chiaroscuro"
+        - "neon glow"
+        - "moonlit"
+        - "backlit silhouette"
+    - name: "Details"
+      desc: "Additional specifications"
+      examples:
+        - "shallow depth of field"
+        - "wide angle lens"
+        - "bird's eye view"
+        - "macro close-up"
+
+text_rendering:
+  description: "Render text inside generated images"
+  method: "Wrap desired text in quotes within prompt"
+  example: "A neon sign that says 'Open 24 Hours' hanging in a rainy alley"
+  tips:
+    - "Use Pro model for best text accuracy"
+    - "Keep text short (1-5 words best)"
+    - "Specify font style: 'bold', 'handwritten', 'neon', 'serif'"
+    - "Describe text placement: 'centered', 'at the bottom', 'on the sign'"
+
+reference_image_patterns:
+  style_transfer:
+    desc: "Apply art style from reference to new content"
+    max_refs: 3
+    prompt: "Generate an image in the same style as the reference: [content description]"
+
+  subject_consistency:
+    desc: "Keep same character/product across images"
+    max_refs: 1
+    prompt: "Generate this character [reference] in a new scene: [scene description]"
+    tip: "Describe the subject's key features in text too for best consistency"
+
+  multi_reference:
+    desc: "Combine elements from multiple references"
+    max_refs: 14 # Pro only
+    prompt: "Combine the style of [ref1], subject from [ref2], background from [ref3]"
+
+  inpainting:
+    desc: "Edit specific region of existing image"
+    requires: "Mask (white = edit, black = keep)"
+    prompt: "Replace [masked area] with [new content]"
+
+style_keywords:
+  photographic:
+    - "35mm film photography"
+    - "DSLR, f/1.4 bokeh"
+    - "polaroid instant photo"
+    - "long exposure light trails"
+    - "infrared photography"
+  artistic:
+    - "impressionist oil painting"
+    - "Japanese woodblock print (ukiyo-e)"
+    - "Art Nouveau poster"
+    - "pop art halftone"
+    - "baroque still life"
+  digital:
+    - "Unreal Engine 5 render"
+    - "octane render, 8K"
+    - "low poly 3D model"
+    - "vaporwave aesthetic"
+    - "cyberpunk neon city"
+
+quality_enhancers:
+  - "highly detailed"
+  - "sharp focus"
+  - "professional lighting"
+  - "award-winning photography"
+  - "4K resolution" # Use with Pro model
+
+negative_prompts:
+  description: "What to AVOID in the image (Nano Banana supports)"
+  examples:
+    - "blurry, out of focus"
+    - "watermark, signature"
+    - "deformed, distorted"
+    - "low quality, pixelated"
+
+vietnamese_support:
+  description: "Gemini supports Vietnamese prompts natively"
+  example: "Một cô gái Việt Nam mặc áo dài trắng đang đi dưới hàng cây phượng vĩ đỏ"
+  tip: "Vietnamese prompts produce culturally accurate imagery"
+
+# =============================================================================
+# _DOMYH Awesome Code • Image Prompt Patterns v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-media-gen/data/video-generation.yaml

@@ -0,0 +1,131 @@
+# =============================================================================
+# video-generation.yaml — Veo 3.1 Video Generation Patterns
+# =============================================================================
+# 10 patterns for text/image-to-video, prompting, extension, frame control
+# Source: https://ai.google.dev/gemini-api/docs/video
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 10
+
+model:
+  id: "veo-3.1"
+  capabilities:
+    - "Text-to-video"
+    - "Image-to-video"
+    - "Video extension"
+    - "Frame control (first/last)"
+    - "Native audio generation"
+    - "Negative prompts"
+  specs:
+    duration: "5-8 seconds"
+    resolutions: ["720p", "1080p", "4K"]
+    audio: "Native AI-generated (sound effects, dialogue, ambient)"
+    aspect_ratios: ["16:9", "9:16"]
+
+prompt_elements:
+  - name: "Subject"
+    desc: "Who/what is in the scene"
+    example: "A chef in a modern kitchen"
+  - name: "Action"
+    desc: "What is happening (be specific about motion)"
+    example: "flipping a pancake high into the air"
+  - name: "Style"
+    desc: "Visual aesthetic"
+    examples:
+      - "cinematic, 35mm film"
+      - "documentary style"
+      - "anime"
+      - "slow motion"
+      - "time-lapse"
+      - "noir black and white"
+  - name: "Camera"
+    desc: "Camera movement and angle"
+    examples:
+      - "pan left slowly"
+      - "zoom in on the subject"
+      - "tracking shot following the character"
+      - "overhead drone shot"
+      - "handheld shaky cam"
+      - "dolly zoom (Vertigo effect)"
+  - name: "Audio"
+    desc: "Sound elements (Veo 3.1 generates native audio)"
+    examples:
+      - "birds chirping in the background"
+      - "dramatic orchestral music"
+      - "the character says 'Welcome home'"
+      - "rain on a tin roof"
+      - "busy city traffic sounds"
+  - name: "Lighting"
+    desc: "Light quality and mood"
+    examples:
+      - "golden hour warm light"
+      - "harsh noon shadows"
+      - "soft studio lighting"
+      - "neon-lit urban night"
+
+generation_modes:
+  text_to_video:
+    desc: "Generate video from text description"
+    api: |
+      response = client.models.generate_videos(
+          model="veo-3.1",
+          prompt="A cat walking through a garden at sunset",
+          config={"generate_audio": True, "aspect_ratio": "16:9"}
+      )
+    tips:
+      - "Be specific about motion and timing"
+      - "Include audio cues for better results"
+      - "Describe camera movement explicitly"
+
+  image_to_video:
+    desc: "Animate a static image"
+    api: |
+      response = client.models.generate_videos(
+          model="veo-3.1",
+          prompt="The flowers sway gently in the breeze",
+          image=reference_image,
+          config={"generate_audio": True}
+      )
+    tips:
+      - "Image becomes the first frame"
+      - "Describe the motion that should happen"
+      - "Works great for product animations"
+
+video_extension:
+  desc: "Extend existing video by generating continuation"
+  how: "Feed last frame of video as reference, prompt describes continuation"
+  max_consecutive: "Multiple extensions possible (each 5-8s)"
+  tip: "Maintain style consistency in extension prompt"
+
+frame_control:
+  first_frame:
+    desc: "Control opening frame with reference image"
+    use_case: "Start from specific composition"
+  last_frame:
+    desc: "Control ending frame with reference image"
+    use_case: "Create transitions between videos"
+
+negative_prompts:
+  desc: "Specify what should NOT appear in video"
+  config_key: "negative_prompt"
+  examples:
+    - "blurry, shaky footage"
+    - "text overlays, watermarks"
+    - "low quality, pixelated"
+
+async_generation:
+  desc: "Video generation is asynchronous (check status)"
+  workflow:
+    - "Submit generation request"
+    - "Receive operation name"
+    - "Poll for completion"
+    - "Download generated video"
+  polling: |
+    while not operation.done:
+        operation = client.operations.get(name=operation.name)
+        time.sleep(5)
+
+# =============================================================================
+# _DOMYH Awesome Code • Video Generation v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-tts/META.yaml

@@ -0,0 +1,49 @@
+name: gemini-tts
+version: "1.0.0"
+display: Gemini Text-to-Speech
+category: ai-ml
+tier: 1
+priority: 3
+
+desc: "Gemini TTS: 3 models (Flash/Lite/Pro), markup tags ([sigh], [whispering], [extremely fast]), multi-speaker, 6 output formats, streaming. Style control via prompts. 29+ patterns."
+
+triggers:
+  file_patterns: []
+  keywords:
+    - text to speech
+    - tts
+    - gemini tts
+    - speech synthesis
+    - voice generation
+    - audio generation
+    - read aloud
+    - narration
+    - multi speaker
+    - voice style
+    - speech markup
+    - prosody
+  intents:
+    - generate speech
+    - convert text to audio
+    - narrate content
+    - create voiceover
+
+caps:
+  - Flash TTS (low-latency streaming)
+  - Lite TTS (cost-effective)
+  - Pro TTS (highest quality)
+  - Markup tags (sigh, laughing, whispering, sarcasm, pauses)
+  - Multi-speaker synthesis
+  - Streaming audio (PCM, OGG_OPUS)
+  - Style control via natural language prompts
+  - Pace & pronunciation control
+  - 6 output audio formats
+
+related_skills:
+  - prompt-engineering
+  - gemini-media-gen
+
+data_files:
+  - data/models.yaml
+  - data/markup-tags.yaml
+  - data/prompting-patterns.yaml

package/.agent/skills/ai-ml/gemini-tts/SKILL.md

@@ -0,0 +1,124 @@
+# Gemini Text-to-Speech
+
+> 🎙️ **Prompt-steerable speech synthesis**
+> **Models**: Flash / Lite / Pro | **Markup tags**: 4 modes | **Patterns**: 29+
+
+---
+
+## Quick Reference
+
+| What You Need              | Data File                 | Patterns |
+| -------------------------- | ------------------------- | -------- |
+| Model comparison & configs | `models.yaml`             | 6        |
+| Markup tags reference      | `markup-tags.yaml`        | 15       |
+| Style prompting            | `prompting-patterns.yaml` | 8        |
+
+---
+
+## 3 Levers of Speech Control
+
+```
+┌──────────────────────────────────────┐
+│  1. Prompt Style Instructions        │  ← Natural language direction
+│  2. Markup Tags in Text              │  ← Fine-grained control
+│  3. Model Selection                  │  ← Quality vs latency
+└──────────────────────────────────────┘
+```
+
+---
+
+## Model Comparison
+
+| Feature           | Flash     | Lite           | Pro                 |
+| ----------------- | --------- | -------------- | ------------------- |
+| **Latency**       | ⚡ Lowest | ⚡ Low         | 🐢 Higher           |
+| **Quality**       | Good      | Good           | **Best**            |
+| **Cost**          | Low       | **Lowest**     | Higher              |
+| **Streaming**     | ✅        | ✅             | ✅                  |
+| **Multi-speaker** | ✅        | ✅             | ✅                  |
+| **Best for**      | Real-time | Cost-effective | Audiobooks, premium |
+
+---
+
+## Markup Tags (4 Modes)
+
+### Non-speech sounds
+
+```
+[laughing] I can't believe that happened!
+[sigh] It's been a long day.
+[gasp] Really?!
+[clears throat] Let me start over.
+```
+
+### Speaking style
+
+```
+[whispering] This is a secret
+[sarcasm] Oh, what a great idea
+[cheerfully] Welcome to the show!
+[sadly] I'm sorry to hear that
+```
+
+### Vocalized sounds
+
+```
+[singing] La la la...
+[humming] Hmm hmm hmm
+```
+
+### Pacing & pronunciation
+
+```
+[extremely fast] Breaking news just in!
+[slowly] Let me explain this carefully
+[pause: 2s] ...and then it happened
+```
+
+---
+
+## Multi-Speaker Setup
+
+```python
+config = {
+    "speech_config": {
+        "multi_speaker_voice_config": {
+            "speaker_voice_configs": [
+                {"speaker": "Alice", "voice_config": {"prebuilt_voice_config": {"voice_name": "Kore"}}},
+                {"speaker": "Bob", "voice_config": {"prebuilt_voice_config": {"voice_name": "Puck"}}}
+            ]
+        }
+    }
+}
+```
+
+Format text with speaker tags:
+
+```
+Alice: Welcome to the podcast!
+Bob: Thanks for having me.
+Alice: [cheerfully] Let's get started!
+```
+
+---
+
+## Output Formats
+
+| Format          | MIME Type               | Use Case                 |
+| --------------- | ----------------------- | ------------------------ |
+| WAV (PCM 24kHz) | `audio/wav`             | High quality, editing    |
+| OGG Opus        | `audio/ogg;codecs=opus` | Streaming, low bandwidth |
+| MP3             | `audio/mp3`             | Universal compatibility  |
+| FLAC            | `audio/flac`            | Lossless archival        |
+| AAC             | `audio/aac`             | Mobile playback          |
+| PCM Raw         | `audio/L16;rate=24000`  | Real-time processing     |
+
+---
+
+## HSA Integration
+
+| Domain | Query Examples                           |
+| ------ | ---------------------------------------- |
+| Models | "TTS model comparison latency quality"   |
+| Markup | "speech markup whispering laughing sigh" |
+| Style  | "energetic narration podcast voiceover"  |