@nockdev/awf 6.2.0 → 6.2.2

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"  |

package/.agent/skills/ai-ml/gemini-tts/data/markup-tags.yaml

@@ -0,0 +1,95 @@
+# =============================================================================
+# markup-tags.yaml — TTS Markup Tags Reference
+# =============================================================================
+# 15 patterns for speech control via inline markup tags
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 15
+
+modes:
+  non_speech_sounds:
+    description: "Add natural non-verbal sounds"
+    tags:
+      - tag: "[laughing]"
+        desc: "Laughter"
+        example: "[laughing] That's the funniest thing I've heard all day!"
+      - tag: "[sigh]"
+        desc: "Exasperated or tired sigh"
+        example: "[sigh] I suppose we'll have to start over."
+      - tag: "[gasp]"
+        desc: "Surprise or shock"
+        example: "[gasp] I had no idea!"
+      - tag: "[clears throat]"
+        desc: "Throat clearing"
+        example: "[clears throat] Allow me to begin."
+      - tag: "[sniffles]"
+        desc: "Sniffling"
+        example: "[sniffles] It's just so beautiful."
+
+  speaking_style:
+    description: "Modify vocal delivery style"
+    tags:
+      - tag: "[whispering]"
+        desc: "Quiet, secretive tone"
+        example: "[whispering] I have something important to tell you..."
+      - tag: "[sarcasm]"
+        desc: "Sarcastic delivery"
+        example: "[sarcasm] Oh wow, what a brilliant plan."
+      - tag: "[cheerfully]"
+        desc: "Happy, upbeat tone"
+        example: "[cheerfully] Great news everyone!"
+      - tag: "[sadly]"
+        desc: "Somber, melancholic tone"
+        example: "[sadly] I'm afraid it's too late."
+      - tag: "[excitedly]"
+        desc: "Energetic, enthusiastic"
+        example: "[excitedly] You won't believe what happened!"
+      - tag: "[calmly]"
+        desc: "Peaceful, measured delivery"
+        example: "[calmly] Take a deep breath and relax."
+
+  vocalized_sounds:
+    description: "Musical and vocalized expressions"
+    tags:
+      - tag: "[singing]"
+        desc: "Singing voice"
+        example: "[singing] Happy birthday to you..."
+      - tag: "[humming]"
+        desc: "Humming melody"
+        example: "[humming] Hmm hmm hmm..."
+
+  pacing_pronunciation:
+    description: "Control speech speed and timing"
+    tags:
+      - tag: "[extremely fast]"
+        desc: "Rapid speech delivery"
+        example: "[extremely fast] And the winner of tonight's grand prize is..."
+      - tag: "[slowly]"
+        desc: "Deliberate, slow delivery"
+        example: "[slowly] Let me explain this one more time..."
+      - tag: "[pause: Xs]"
+        desc: "Insert silence (X seconds)"
+        example: "The answer is... [pause: 2s] ...forty-two."
+
+usage_tips:
+  - "Tags go BEFORE the text they modify"
+  - "Tags affect the sentence/phrase immediately following"
+  - "Combine tags: '[whispering][slowly] This is very important...'"
+  - "Tags are case-insensitive: [LAUGHING] = [laughing]"
+  - "Not all tags work equally well — experiment with model"
+  - "Pro model handles markup most naturally"
+
+combinations:
+  example_dialogue: |
+    [cheerfully] Welcome to today's episode!
+    [pause: 1s]
+    [calmly] Today we're going to discuss something fascinating.
+    [excitedly] You won't believe the results we found!
+    [whispering] But first, let me share a little secret...
+    [laughing] Just kidding, it's not that dramatic.
+    [sigh] Anyway, let's get started.
+
+# =============================================================================
+# _DOMYH Awesome Code • Markup Tags v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-tts/data/models.yaml

@@ -0,0 +1,124 @@
+# =============================================================================
+# models.yaml — Gemini TTS Model Comparison
+# =============================================================================
+# 6 patterns for model selection, voice options, output configuration
+# Source: https://docs.cloud.google.com/text-to-speech/docs/gemini-tts
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 6
+
+models:
+  flash:
+    id: "gemini-2.5-flash-preview-tts"
+    display: "Gemini TTS Flash"
+    tier: "speed"
+    latency: "lowest"
+    quality: "good"
+    cost: "low"
+    streaming: true
+    multi_speaker: true
+    best_for:
+      - "Real-time applications"
+      - "Interactive voice agents"
+      - "Live narration"
+
+  lite:
+    id: "gemini-2.5-flash-lite-preview-tts"
+    display: "Gemini TTS Lite"
+    tier: "cost"
+    latency: "low"
+    quality: "good"
+    cost: "lowest"
+    streaming: true
+    multi_speaker: true
+    best_for:
+      - "High-volume batch processing"
+      - "Cost-sensitive applications"
+      - "Notifications and alerts"
+
+  pro:
+    id: "gemini-2.5-pro-preview-tts"
+    display: "Gemini TTS Pro"
+    tier: "quality"
+    latency: "higher"
+    quality: "best"
+    cost: "higher"
+    streaming: true
+    multi_speaker: true
+    best_for:
+      - "Audiobooks"
+      - "Premium voiceovers"
+      - "Marketing content"
+      - "Podcast production"
+
+voices:
+  description: "Prebuilt voice options available across all models"
+  options:
+    - name: "Kore"
+      style: "Clear, professional female"
+    - name: "Puck"
+      style: "Warm, conversational male"
+    - name: "Charon"
+      style: "Deep, authoritative male"
+    - name: "Aoede"
+      style: "Bright, energetic female"
+    - name: "Fenrir"
+      style: "Calm, measured male"
+    - name: "Leda"
+      style: "Friendly, approachable female"
+
+output_formats:
+  - mime: "audio/wav"
+    desc: "WAV (PCM 24kHz)"
+    use_case: "High quality editing"
+  - mime: "audio/ogg;codecs=opus"
+    desc: "OGG Opus"
+    use_case: "Streaming, low bandwidth"
+  - mime: "audio/mp3"
+    desc: "MP3"
+    use_case: "Universal compatibility"
+  - mime: "audio/flac"
+    desc: "FLAC"
+    use_case: "Lossless archival"
+  - mime: "audio/aac"
+    desc: "AAC"
+    use_case: "Mobile playback"
+  - mime: "audio/L16;rate=24000"
+    desc: "Raw PCM 24kHz"
+    use_case: "Real-time processing"
+
+api_usage:
+  basic:
+    python: |
+      from google import genai
+      client = genai.Client(api_key="API_KEY")
+      response = client.models.generate_content(
+          model="gemini-2.5-flash-preview-tts",
+          contents="Hello, welcome to our product demo!",
+          config=genai.types.GenerateContentConfig(
+              response_modalities=["AUDIO"],
+              speech_config=genai.types.SpeechConfig(
+                  voice_config=genai.types.VoiceConfig(
+                      prebuilt_voice_config=genai.types.PrebuiltVoiceConfig(
+                          voice_name="Kore"
+                      )
+                  )
+              )
+          )
+      )
+      # Save: response.candidates[0].content.parts[0].inline_data.data
+
+  streaming:
+    python: |
+      for chunk in client.models.generate_content_stream(
+          model="gemini-2.5-flash-preview-tts",
+          contents="Long text to narrate...",
+          config=config
+      ):
+          # Process each audio chunk in real-time
+          audio_data = chunk.candidates[0].content.parts[0].inline_data.data
+
+# =============================================================================
+# _DOMYH Awesome Code • TTS Models v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/gemini-tts/data/prompting-patterns.yaml

@@ -0,0 +1,81 @@
+# =============================================================================
+# prompting-patterns.yaml — TTS Style Prompting Patterns
+# =============================================================================
+# 8 patterns for controlling TTS output via natural language prompts
+# =============================================================================
+
+version: "1.0.0"
+total_patterns: 8
+
+style_prompts:
+  description: "Use natural language to direct voice style"
+  principle: "System instruction sets the voice persona and delivery style"
+
+  patterns:
+    news_anchor:
+      prompt: "Read this in a professional, authoritative news anchor voice with clear enunciation."
+      use_case: "News summaries, announcements"
+
+    podcast_host:
+      prompt: "Read this in a warm, conversational podcast host voice. Be engaging and natural."
+      use_case: "Podcast scripts, casual narration"
+
+    storyteller:
+      prompt: "Read this like a bedtime story narrator — gentle, expressive, with dramatic pauses."
+      use_case: "Audiobooks, children's content"
+
+    instructor:
+      prompt: "Read this like a patient, clear teacher explaining a concept. Emphasize key terms."
+      use_case: "Educational content, tutorials"
+
+    dramatic:
+      prompt: "Read this with dramatic flair, varying pace and intensity for emotional impact."
+      use_case: "Theater scripts, dramatic readings"
+
+    meditation:
+      prompt: "Read this in a calm, soothing meditation guide voice. Slow pace, soft tone."
+      use_case: "Guided meditation, relaxation content"
+
+    sports_commentator:
+      prompt: "Read this with the energy and excitement of a live sports commentator."
+      use_case: "Sports updates, exciting announcements"
+
+    documentary:
+      prompt: "Read this in a BBC documentary narrator style — informative, measured, with gravitas."
+      use_case: "Documentary narration, educational videos"
+
+emotionally_rich_text:
+  description: "Write text that naturally triggers expressive delivery"
+  tips:
+    - "Use punctuation for natural pauses: commas, ellipses, em dashes"
+    - "Exclamation marks increase energy: 'Amazing!' vs 'Amazing.'"
+    - "Question marks trigger interrogative intonation"
+    - "Longer sentences → more natural flow"
+    - "Short sentences → punchy, impactful delivery"
+
+  example_text: |
+    Have you ever wondered... what makes a great voice?
+    It's not just about clarity — it's about emotion.
+    The way a pause... can build anticipation.
+    The way excitement can make your heart race!
+    And sometimes, a whisper is the most powerful tool of all.
+
+alignment_strategies:
+  description: "Ensure voice style matches content tone"
+  rules:
+    - "Match prompt style to content genre (don't use news voice for poetry)"
+    - "Use markup tags for fine-grained moments within a consistent style"
+    - "Test with short segments before full narration"
+    - "Pro model best for style adherence"
+    - "Multi-speaker: assign different styles per speaker"
+
+multi_language:
+  description: "TTS supports multiple languages"
+  tips:
+    - "Specify language in prompt: 'Read in Vietnamese with natural pronunciation'"
+    - "Model auto-detects language from text"
+    - "Code-switching (mixed languages) works but quality varies"
+
+# =============================================================================
+# _DOMYH Awesome Code • TTS Prompting Patterns v1.0_
+# =============================================================================

package/.agent/skills/ai-ml/prompt-engineering/META.yaml

@@ -0,0 +1,77 @@
+name: prompt-engineering
+version: "2.0.0"
+display: Prompt Engineering Patterns
+category: ai-ml
+tier: 1
+priority: 1
+
+desc: "Prompt engineering for LLMs: chain-of-thought, few-shot, ReAct, tree-of-thought, structured output, system prompts, guardrails. Gemini 3 agentic 9-rule template, thinking controls, function calling, MCP. Multi-provider patterns (OpenAI, Anthropic, Google). 155+ patterns."
+
+triggers:
+  file_patterns: []
+  keywords:
+    - prompt
+    - prompt engineering
+    - chain of thought
+    - few-shot
+    - zero-shot
+    - system prompt
+    - llm
+    - ai
+    - instruction
+    - reasoning
+    - react pattern
+    - tree of thought
+    - structured output
+    - guardrail
+    - jailbreak
+    # Gemini 3 / v2.0
+    - gemini 3
+    - agentic template
+    - thinking
+    - thinking budget
+    - function calling
+    - compositional function
+    - MCP function calling
+    - grounding
+    - context caching
+  intents:
+    - write prompts
+    - prompt design
+    - LLM integration
+    - AI instruction
+    - agentic system design
+
+caps:
+  - Chain-of-thought (CoT) prompting
+  - Few-shot / zero-shot learning
+  - ReAct (Reasoning + Acting)
+  - Tree-of-thought (ToT)
+  - Structured output (JSON mode, Pydantic, Zod)
+  - System prompt design
+  - Prompt injection defense
+  - Multi-turn conversation design
+  - Evaluation and testing prompts
+  - Tool use / function calling
+  - RAG prompt patterns
+  - Multi-provider patterns (OpenAI, Anthropic, Google)
+  # v2.0
+  - Gemini 3 agentic 9-rule template
+  - Thinking controls (levels, budgets, summaries)
+  - Function calling (parallel, compositional, MCP)
+  - Context caching strategies
+  - Grounding with Google Search
+
+related_skills:
+  - ai-agents
+  - gemini-live
+  - gemini-media-gen
+
+data_files:
+  - data/reasoning-patterns.yaml
+  - data/output-patterns.yaml
+  - data/safety-patterns.yaml
+  - data/provider-patterns.yaml
+  # v2.0
+  - data/gemini3-patterns.yaml
+  - data/tool-patterns.yaml