### Backend Setup with npm

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/convex-swift-quickstart.md

Steps to initialize a Node.js project, install the Convex package, and start the Convex development server.

```bash
# In project root directory
npm init -y
npm install convex

# Start Convex development server
npx convex dev
```

--------------------------------

### High-level Guide with References in SKILL.md

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills.md

Demonstrates a pattern for organizing a SKILL.md file to provide a quick start guide and link to detailed supporting files.

```markdown
# PDF Processing

## Quick start

Extract text with pdfplumber:
```python
import pdfplumber
with pdfplumber.open("file.pdf") as pdf:
    text = pdf.pages[0].extract_text()
```

## Advanced features

**Form filling**: See [FORMS.md](FORMS.md)
**API reference**: See [REFERENCE.md](REFERENCE.md)
**Examples**: See [EXAMPLES.md](EXAMPLES.md)

```

--------------------------------

### Parallel Research Execution Example

Source: https://github.com/captaincrouton89/.claude/blob/main/output-styles/deep-research.md

This example demonstrates initiating multiple parallel research agents for distinct aspects of authentication, including implementation details, security audits, and best practices research. It shows the commands to start these agents and then synchronize their results.

```bash
klaude start Explore "Deep dive into authentication implementation:

Priority Questions:
- Token generation and validation mechanisms
- Session management lifecycle
- Security vulnerability patterns

Requirements:
- Read all files in auth/ directory
- Trace authentication flow from entry to validation
- Document security measures with confidence ratings
- Note any potential vulnerabilities or gaps

Output: Detailed implementation report with code references"
```

```bash
klaude start Explore "Security audit of authentication system:

Focus Areas:
- OWASP compliance check
- Rate limiting implementation
- Token security (storage, transmission, rotation)
- Injection vulnerability analysis

Requirements:
- Search for security anti-patterns
- Verify all user inputs are sanitized
- Check for timing attacks in validation
- Document findings with CVE references where applicable

Output: Security assessment with risk ratings"
```

```bash
klaude start general-purpose "Research authentication best practices and patterns:

Investigation targets:
- Industry standard implementations (OAuth2, JWT, SAML)
- Current security recommendations (2024-2025)
- Framework-specific patterns for our stack

Requirements:
- Use WebSearch for latest security advisories
- Find authoritative sources (IETF, OWASP, framework docs)
- Compare our implementation to standards
- Note deviations with justification analysis

Output: Standards comparison with recommendations"
```

```bash
klaude wait <impl-session-id> <security-session-id> <patterns-session-id>
```

--------------------------------

### Quick Start: Gemini API Text and Image Generation

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/gemini.md

Basic example showing API key configuration, model initialization, text generation, and image analysis using the Gemini API.

```python
import google.generativeai as genai
import os

# Configure API key
genai.configure(api_key=os.environ["GOOGLE_AI_API_KEY"])

# Initialize model
model = genai.GenerativeModel("gemini-2.5-flash")

# Text generation
response = model.generate_content("Explain quantum computing")
print(response.text)

# Image analysis
image_file = genai.upload_file("photo.jpg")
response = model.generate_content([
    "What's happening in this image?",
    image_file
])
print(response.text)
```

--------------------------------

### Documentation Generation Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/CLAUDE.md

Example of exporting project documentation to markdown format.

```bash
./docs/generate-docs.sh
```

--------------------------------

### Install Cursor CLI

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/cursor-cli-reference.md

Use this command to download and install the Cursor CLI.

```bash
curl https://cursor.com/install -fsSL | bash
```

--------------------------------

### API Documentation Generation Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/CLAUDE.md

Example of generating API documentation in curl format using the list-apis script.

```bash
./docs/list-apis.sh --format curl
```

--------------------------------

### Observer Trigger and Workflow Examples

Source: https://github.com/captaincrouton89/.claude/blob/main/multiagent/ObserverPattern.md

Examples of triggers, workflows, and notifications for different meta observers.

```text
Trigger on: vague responses, conflicting directions, "wait what?"
Workflow: Clarification questions, assumption validation
Notification: "Let me make sure I understand correctly..."
```

```text
Trigger on: todo list growing rapidly, new requirements mid-task
Workflow: Task decomposition, priority assessment
Notification: "This is growing - let me help break it down."
```

```text
Trigger on: "should", "probably", "I think", unverified claims
Workflow: Assumption extraction, validation research
Notification: "Checking some assumptions here."
```

```text
Trigger on: code duplication, similar patterns in different files
Workflow: Refactoring suggestions, DRY opportunities
Notification: "I noticed some refactoring opportunities."
```

```text
Trigger on: complex code, API changes, new patterns
Workflow: Doc generation, README updates
Notification: "Updating documentation for this."
```

```text
Trigger on: new code that differs from existing patterns
Workflow: Pattern comparison, style guide validation
Notification: "Checking consistency with existing patterns."
```

--------------------------------

### Common Usage Pattern Example

Source: https://github.com/captaincrouton89/.claude/blob/main/file-templates/feature-doc.template.md

Illustrates a typical way to use the feature. This snippet serves as a starting point for understanding the feature's practical application.

```typescript
// Common usage pattern
[code example]
```

--------------------------------

### Example Directory Structure

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/agent-doc.md

Illustrates a recommended documentation structure with a main overview file and a separate reference directory for detailed topics like API, examples, and troubleshooting.

```plaintext
my-doc/
├── OVERVIEW.md           # High-level guide
├── reference/
│   ├── api.md           # Specific reference
│   ├── examples.md      # Usage examples
│   └── troubleshooting.md
└── scripts/
    └── helper.py        # Executable utilities
```

--------------------------------

### Project Validation Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/CLAUDE.md

Example of validating all project documentation files using the check-project script with verbose output.

```bash
./docs/check-project.sh -v
```

--------------------------------

### Quick Start: Gemini API Video Analysis

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/gemini.md

Example demonstrating video analysis using the Gemini API, including file upload and content generation request.

```python
video_file = genai.upload_file("video.mp4")
response = model.generate_content([
    "Summarize the key events in this video",
    video_file
])
print(response.text)
```

--------------------------------

### Load All Filesystem Settings

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/typescript-sdk.md

Example demonstrating how to load all filesystem settings, mimicking the behavior of SDK v0.0.x.

```typescript
// Load all settings like SDK v0.0.x did
const result = query({
  prompt: "Analyze this code",
  options: {
    settingSources: ['user', 'project', 'local']  // Load all settings
  }
});
```

--------------------------------

### Node.js Text-to-Speech Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/elevenlabs.md

Generate speech from text using the ElevenLabs Node.js SDK. Ensure you have the SDK installed (`npm install @elevenlabs/elevenlabs-js`) and replace 'YOUR_API_KEY' with your actual API key.

```JavaScript
import { ElevenLabsClient } from '@elevenlabs/elevenlabs-js';

const client = new ElevenLabsClient({
  apiKey: 'YOUR_API_KEY',
});

// Text to Speech
const audio = await client.generate({
  text: "Hello, world!",
  voice: "Rachel",
  model_id: "eleven_multilingual_v2"
});
```

--------------------------------

### Python Text-to-Speech Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/elevenlabs.md

Generate speech from text using the ElevenLabs Python SDK. Ensure you have the SDK installed (`pip install elevenlabs`) and replace 'YOUR_API_KEY' with your actual API key.

```Python
from elevenlabs.client import ElevenLabs

client = ElevenLabs(api_key='YOUR_API_KEY')

# Text to Speech
audio = client.generate(
    text="Hello, world!",
    voice="Rachel",
    model="eleven_multilingual_v2"
)
```

--------------------------------

### Install FFmpeg on Linux (Ubuntu/Debian)

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/ffmpeg-analysis.md

Installs FFmpeg using the apt package manager.

```bash
sudo apt update
sudo apt install ffmpeg
```

--------------------------------

### Provide Aligned Example for Commit Message

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/prompting-effectively/SKILL.md

Use `<example>` tags to provide a concrete input-output pair for generating commit messages, demonstrating the desired format and content.

```plaintext
<example>
Input: "Added JWT authentication"
Output:
feat(auth): implement JWT-based authentication

Add login endpoint and token validation middleware.
</example>
```

--------------------------------

### Install TypeScript SDK

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/features/typescript-sdk-reference.doc.md

Installs the TypeScript Agent SDK using npm. Ensure you have Node.js and npm installed.

```bash
npm install @anthropic-ai/claude-agent-sdk
```

--------------------------------

### Good Description Example for SKILL.md

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills.md

Provides a concrete example of a good description for the SKILL.md frontmatter, detailing capabilities and usage context.

```yaml
description: Extract text and tables from PDF files, fill forms, merge documents.
Use when working with PDF files or when the user mentions PDFs, forms, or extraction.

```

--------------------------------

### Good Description Example

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/skills-guide/SKILL.md

This example demonstrates a specific and actionable description for a skill, including its capabilities and usage context.

```text
Extract text and tables from PDF files, fill forms, merge documents.
Use when working with PDF files or when the user mentions PDFs, forms, or extraction.
```

--------------------------------

### Install OpenAI Whisper

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/whisper-transcription.md

Install the Whisper Python package. Use '[complete]' for advanced features.

```bash
pip install openai-whisper
```

```bash
pip install openai-whisper[complete]
```

--------------------------------

### Simple Prompt Shortcut Example

Source: https://github.com/captaincrouton89/.claude/blob/main/workflows/custom-slash-commands-guide.md

A straightforward command that takes arguments and analyzes them for specific security vulnerabilities.

```markdown
---
description: Review code for security issues
---

Analyze $ARGUMENTS for security vulnerabilities including:
- SQL injection
- XSS attacks
- Authentication issues
- Data exposure
```

--------------------------------

### Chaining Subagents Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/agents.md

Example of chaining subagents for complex workflows. This demonstrates how to use the code-analyzer and optimizer agents sequentially to address performance issues.

```bash
> First use the code-analyzer subagent to find performance issues, then use the optimizer subagent to fix them
```

--------------------------------

### Read Tool Input Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Input for the Read tool, specifying the file path to read.

```json
{
  "file_path": "~/Code/Klaude/test.txt"
}
```

--------------------------------

### Example of Specific Skill Descriptions

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/skills-guide/SKILL.md

YAML examples demonstrating how to write specific and trigger-aware descriptions for Claude skills to avoid conflicts and ensure correct activation.

```yaml
# Skill 1
description: Analyze sales data in Excel and CRM exports.
Use for sales reports, pipeline analysis, and revenue tracking.

# Skill 2
description: Analyze log files and system metrics.
Use for performance monitoring, debugging, and system diagnostics.
```

--------------------------------

### Observer Trigger Examples

Source: https://github.com/captaincrouton89/.claude/blob/main/multiagent/ObserverPattern.md

Provides examples of triggers and corresponding workflows/notifications for different observer categories. These illustrate the specific conditions under which observers activate and the actions they initiate.

```text
Creativity Observer:
Trigger on: stuck, repetition, "I need ideas"
Workflow: Lateral thinking, analogies, alternative approaches
Notification: "I'll brainstorm some creative angles on this."

Deep Research Observer:
Trigger on: complex questions, "how does X work", library names
Workflow: Multi-source research, web search, doc aggregation
Notification: "I'm researching this more deeply - one moment."

Learning Observer:
Trigger on: "I don't understand", confusion signals, repeated questions
Workflow: ELI5 explanations, examples, interactive teaching
Notification: "Let me explain this more clearly."

Security Observer:
Trigger on: auth code, API keys, user input handling, file operations
Workflow: Security analysis, vulnerability scan, best practices check
Notification: "Running security analysis on this code."

Performance Observer:
Trigger on: loops in code, database queries, "it's slow"
Workflow: Profiling suggestions, optimization opportunities
Notification: "Analyzing performance implications."

Testing Observer:
Trigger on: new feature completed, complex logic, critical path code
Workflow: Test generation, edge case identification
Notification: "Generating tests for this."
```

--------------------------------

### Install and Authenticate Wrangler CLI

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/r2-image-hosting.md

Installs the Wrangler CLI globally and authenticates your account with Cloudflare. This is the first step for managing R2 resources via the command line.

```bash
# Install globally
npm install -g wrangler

# Or use pnpm
pnpm add -g wrangler

# Authenticate with Cloudflare
wrangler login
```

--------------------------------

### Write Tool Input Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Input for the Write tool, specifying the file path and content to write.

```json
{
  "content": "test",
  "file_path": "~/Code/Klaude/test.txt"
}
```

--------------------------------

### Vertex AI API Endpoint Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/gemini-video-analysis.md

Example of a POST request to the Vertex AI API for content generation, including project, location, and model parameters.

```bash
POST https://aiplatform.googleapis.com/v1/projects/{project}/locations/{location}/publishers/google/models/{model}:generateContent
```

--------------------------------

### TodoWrite Tool Output Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Output from the TodoWrite tool, showing the updated state of todos.

```json
{
  "newTodos": [
    {
      "activeForm": "Writing a short test file",
      "content": "Write a short test file",
      "status": "in_progress"
    },
    {
      "activeForm": "Running bash echo command",
      "content": "Run bash echo command",
      "status": "pending"
    }
  ],
  "oldTodos": [
    {
      "activeForm": "Writing a short test file",
      "content": "Write a short test file",
      "status": "pending"
    },
    {
      "activeForm": "Running bash echo command",
      "content": "Run bash echo command",
      "status": "pending"
    }
  ]
}
```

--------------------------------

### Read Tool Output Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Output from the Read tool, containing the content and metadata of the specified file.

```json
{
  "file": {
    "content": "Test",
    "filePath": "~/Code/Klaude/test.txt",
    "numLines": 1,
    "startLine": 1,
    "totalLines": 1
  },
  "type": "text"
}
```

--------------------------------

### Quick Start: Create First Command

Source: https://github.com/captaincrouton89/.claude/blob/main/workflows/custom-slash-commands-guide.md

Demonstrates the bash commands to create the necessary directory and a simple 'review' command file.

```bash
mkdir -p .claude/commands
echo "---
description: Quick code review
---

Review $ARGUMENTS for:
- Code quality
- Best practices
- Potential bugs" > .claude/commands/review.md
```

--------------------------------

### Example Usage of Planning Command

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/features/planning-templates.doc.md

Demonstrates how a user initiates a planning command and the system's response, including template referencing.

```bash
# User runs planning command
/shared

# Command reads template and instructs:
# "make a docs/plans/[plan-dir]/shared.md document...
# using the template ~/.claude/file-templates/shared.template.md"
```

--------------------------------

### Example: Copy Skills to New Project

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills-management.md

Demonstrates copying skills to a new project's local skills directory.

```bash
fetch-skills copy ./.claude/skills
```

--------------------------------

### Loading CLAUDE.md Project Instructions

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/typescript-sdk.md

Example demonstrating how to load project settings to include CLAUDE.md files by specifying the system prompt type and setting sources.

```typescript
// Load project settings to include CLAUDE.md files
const result = query({
  prompt: "Add a new feature following project conventions",
  options: {
    systemPrompt: {
      type: 'preset',
      preset: 'claude_code'  // Required to use CLAUDE.md
    },
    settingSources: ['project'],  // Loads CLAUDE.md from project directory
    allowedTools: ['Read', 'Write', 'Edit']
  }
});
```

--------------------------------

### Progressive Disclosure Example in SKILL.md

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/skills-guide/SKILL.md

Demonstrates how to use progressive disclosure in SKILL.md by providing a quick start and linking to more detailed reference files.

```markdown
## Quick start

Extract text with pdfplumber:
```python
import pdfplumber
with pdfplumber.open("file.pdf") as pdf:
    text = pdf.pages[0].extract_text()
```

## Advanced features

**Form filling**: See [FORMS.md](FORMS.md)
**API reference**: See [REFERENCE.md](REFERENCE.md)
**Examples**: See [EXAMPLES.md](EXAMPLES.md)
```

--------------------------------

### Simple Scene Detection (CLI)

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/opencv-scene-detection.md

Performs basic scene detection on a video file using the command line. No additional setup is required beyond installation.

```bash
scenedetect -i video.mp4 detect-content
```

--------------------------------

### User Prompt Example for Style Switching

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/features/automatic-output-style-switching.doc.md

Demonstrates how user prompts with specific keywords can trigger automatic changes in the output style.

```bash
# User prompt automatically triggers style switching
User: "Let's brainstorm some creative solutions for user onboarding"
# System switches to Brainstorming output style

User: "Now implement the login feature we discussed"
# System switches to Sr. Software Developer output style
```

--------------------------------

### Check Cursor Agent Version and Status

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/cursor-cli-reference.md

Checks the installed version of the Cursor agent and verifies authentication and connection status. Essential for troubleshooting and ensuring proper setup.

```bash
cursor-agent --version
```

```bash
cursor-agent status  # Check auth and connection
```

--------------------------------

### POST /items/{item_id}/upgrade-cost (Initialize Profile)

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/generated/api-reference.md

Initializes a new player profile after registration, creating starter items and gold.

```APIDOC
## POST /items/{item_id}/upgrade-cost

### Description
Initialize new player profile (after registration).
Creates starting inventory (6 level-1 items, 500 gold).

### Method
POST

### Endpoint
/items/{item_id}/upgrade-cost

### Responses
#### Success Response (201)
- Profile created with starter items
```

--------------------------------

### Current Flow Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/agent-session-entry-analysis.md

Illustrates the step-by-step process of entering and exiting an agent session, highlighting the session forking problem.

```text
User in Parent Session S1 (PID 100)
  ↓
Task spawns agent → Agent Session S2 created (PID 200) in registry
  ↓
User runs enter-agent
  - Finds S2 in registry
  - Writes marker: S2
  - Kills Claude PID 100
  - Wrapper detects marker
  ↓
Wrapper runs: claude --resume S2
  - Creates FORKED session S2' (different instance)
  - S2' is interactive in terminal
  ↓
User in agent (S2') now
  ↓
User runs exit-agent
  - Finds parent S1 in registry (via agent entry)
  - Writes marker: S1
  - Kills Claude (managing S2')
  - Wrapper detects marker
  ↓
Wrapper runs: claude --resume S1
  - Creates FORKED session S1' (different instance)
  - Problem: Original Task spawned in S1 is still expecting S1, not S1'
  - Task sees orphaned/forked state
```

--------------------------------

### Install PySceneDetect with OpenCV

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/opencv-scene-detection.md

Installs PySceneDetect along with OpenCV support using pip. Ensure Python 3.7+ and NumPy are installed.

```bash
pip install scenedetect[opencv]
```

--------------------------------

### Install FFmpeg on macOS

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/ffmpeg-analysis.md

Installs FFmpeg using the Homebrew package manager.

```bash
brew install ffmpeg
```

--------------------------------

### Skill Discovery on Project Initialization

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills-management.md

Demonstrates the process of initializing a new project with Claude, where skills are automatically discovered, fetched, and curated based on project type.

```bash
$ cd my-new-project
$ claude /init-workspace

# Hook detects /init-workspace command
# Hook runs: fetch-skills table
# Skill list injected into agent prompt
#
# Agent receives:
# - Available skills table
# - Instructions to copy and curate
# - Guidance on which skills to keep for project type
#
# Agent completes MCP + agent setup, then runs:
# $ fetch-skills copy ./.claude/skills
#
# Agent deletes irrelevant skills, keeps only project-appropriate ones
# Result: my-new-project/.claude/skills/ contains only relevant skills
```

--------------------------------

### Install OpenAI Provider

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/open-ai-provider-vercel-ai-sdk.md

Install the OpenAI provider package using your preferred package manager.

```bash
pnpm add @ai-sdk/openai
```

```bash
npm install @ai-sdk/openai
```

```bash
yarn add @ai-sdk/openai
```

```bash
bun add @ai-sdk/openai
```

--------------------------------

### Analyze Video and Generate Recreation (4-8s Clips)

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/README.md

This workflow outlines the steps for processing short video clips. It involves analyzing video with Gemini, extracting and transcribing audio with ElevenLabs, generating a recreation with Replicate, creating new audio with ElevenLabs TTS, and combining everything using ffmpeg.

```bash
# 1. Analyze video with Gemini
# 2. Extract audio, transcribe with ElevenLabs
# 3. Generate recreation with Replicate Veo 3
# 4. Generate audio with ElevenLabs TTS
# 5. Combine using ffmpeg
```

--------------------------------

### Verify FFmpeg and FFprobe Installation

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/ffmpeg-analysis.md

Checks if FFmpeg and FFprobe are installed and accessible in the system's PATH.

```bash
ffmpeg -version
ffprobe -version
```

--------------------------------

### Gemini API Endpoint Example (Flash)

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/gemini-video-analysis.md

Specific example of a POST request targeting the gemini-2.5-flash model.

```bash
POST https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent
```

--------------------------------

### Install OpenCV

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/opencv-scene-detection.md

Installs the OpenCV library for Python using pip. The optional package provides GUI support.

```bash
pip install opencv-python
```

```bash
pip install opencv-contrib-python
```

--------------------------------

### Image-to-Video Recreation Workflow

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/README.md

This workflow demonstrates how to create videos from images. It involves extracting key frames, generating similar images, animating them, and adding audio.

```bash
# 1. Extract key frames from original
# 2. Generate similar images with Gemini/Imagen
# 3. Animate images with Replicate I2V models
# 4. Add audio with ElevenLabs
```

--------------------------------

### Basic Query with Project Settings

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/features/typescript-sdk-reference.doc.md

Demonstrates a basic query using the `query` function with project settings loaded from CLAUDE.md. It shows how to process messages and results from the SDK.

```typescript
import { query } from '@anthropic-ai/claude-agent-sdk';

// Basic query with project settings
const result = query({
  prompt: "Add a new feature following project conventions",
  options: {
    systemPrompt: {
      type: 'preset',
      preset: 'claude_code'
    },
    settingSources: ['project'],  // Load CLAUDE.md
    allowedTools: ['Read', 'Write', 'Edit', 'Grep', 'Glob']
  }
});

// Process messages
for await (const message of result) {
  if (message.type === 'assistant') {
    console.log('Claude:', message.message);
  } else if (message.type === 'result') {
    console.log('Cost:', message.total_cost_usd);
    console.log('Result:', message.result);
  }
}
```

--------------------------------

### Audio Input with OpenAI gpt-4o-audio-preview

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/open-ai-provider-vercel-ai-sdk.md

Example of passing audio files to the `gpt-4o-audio-preview` model for processing. Requires audio inputs and is currently in preview.

```typescript
import { openai } from '@ai-sdk/openai';
import { generateText } from 'ai';

const result = await generateText({
  model: openai.chat('gpt-4o-audio-preview'),
  messages: [
    {
      role: 'user',
      content:
        [
          { type: 'text', text: 'What is the audio saying?' },
          {
            type: 'file',
            mediaType: 'audio/mpeg',
            data: fs.readFileSync('./data/galileo.mp3'),
          },
        ],
    },
  ],
});
```

--------------------------------

### Hook Configuration Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills-management.md

Shows how to register the auto-copy-skills hook in the settings.json file for automatic skill injection on project initialization.

```json
{
  "UserPromptSubmit": [
    {
      "command": "~/.claude/hooks/user-prompt-submit/auto-copy-skills.mjs",
      "timeout": 5000
    }
  ]
}
```

--------------------------------

### SessionStart Hook Event Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Triggered when a new session begins. Includes current working directory, session ID, and transcript path.

```json
{
  "cwd": "~/Code/Klaude",
  "hook_event_name": "SessionStart",
  "session_id": "10ad1de4-613a-470f-9463-bf4fcac617eb",
  "source": "startup",
  "transcript_path": "~/.claude/projects/-Users-silasrhyneer-Code-Klaude/10ad1de4-613a-470f-9463-bf4fcac617eb.jsonl"
}
```

--------------------------------

### Orchestrator Final Report Example

Source: https://github.com/captaincrouton89/.claude/blob/main/agents-library/orchestrator.md

An example of a final report generated by the orchestrator upon successful task completion.

```text
✅ Task completed successfully

Summary:
- Investigated 4 subsystems
- Created implementation plan (docs/plans/auth/plan.md)
- Implemented 15 refactors across 23 files
- Build validation passed

Agent work:
- Explore: agent_001, agent_002, agent_003, agent_004
- Plan: agent_005
- programmer: agent_006, agent_007
- junior-engineer: agent_008 - agent_015
```

--------------------------------

### Project Directory Structure Example

Source: https://github.com/captaincrouton89/.claude/blob/main/ARCHITECTURE-OVERVIEW.md

Illustrates the hierarchical organization of project directories, including agent definitions, libraries, documentation, cached projects, and state management files.

```tree
agents/
├── Explore.md             # Context discovery specialist (YOURS - search engine)
├── Plan.md                # Planning & architecture specialist
├── general-purpose.md     # Default fallback agent
├── programmer.md          # Full-stack programmer
├── senior-programmer.md   # Senior-level code review
├── senior-architect.md    # Architecture decisions
├── junior-engineer.md     # Guided learning / junior work
├── frontend-engineer.md   # Frontend specialization
├── non-dev.md             # Non-developer tasks
├── advisor.md             # Strategic advising
├── library-docs-writer.md # Documentation specialist
└── statusline-setup.md    # Status line configuration
```

```tree
agents-library/
├── db-modifier.md        # Database schema modifications
├── documentor.md         # Documentation generation
├── orchestrator.md       # Multi-agent orchestration
├── product-designer.md   # Product design workflows
├── research-specialist.md # Research & analysis
└── video/                # Video-related agents
```

```tree
docs/
├── CLAUDE.md                      # Doc directory conventions
├── product-requirements.md        # PRD with features (F-##)
├── system-design.md              # Architecture & components
├── design-spec.md                # UI/UX screens
├── api-contracts.yaml            # OpenAPI definitions
├── data-plan.md                  # Metrics & tracking
├── feature-spec/                 # Feature specs (F-##-*.md)
├── user-stories/                 # User stories (US-###-*.md)
├── user-flows/                   # Primary user flows
├── guides/                        # Process docs & runbooks
├── architecture/                 # Architecture decisions
└── external/                     # External docs & references
```

```tree
projects/
├── -Users-silasrhyneer-Code-claude-tools-klaude/  # Main project
├── -Users-silasrhyneer-Code-Cosmo-Saturn/         # Other projects
└── [path-encoded project directories]/
    ├── .claude/                   # Project-specific state
    ├── state/                     # Session state
    ├── todos/                     # Todo tracking
    ├── debug/                     # Debug logs
    └── file-history/              # File change history
```

```tree
conversation-state/
└── {session_id}.json     # Protocol name, effort level, timestamp tracking
```

```tree
logs/
├── hooks.log     # All hook execution events
└── [other logs]
```

--------------------------------

### Tool Call Started Event

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/cursor-cli-reference.md

This JSON object indicates the start of a tool call in the stream-json output format.

```json
{
  "type": "tool_call",
  "subtype": "started",
  "call_id": "call-uuid",
  "tool_name": "read_file",
  "args": {
    "path": "src/auth.js"
  },
  "session_id": "uuid-here"
}
```

--------------------------------

### Bad Description Example

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/skills-guide/SKILL.md

This example shows a vague description that is not specific enough for Claude to effectively determine when to use the skill.

```text
Helps with documents
```

--------------------------------

### Usage Examples for Workflow Reminders

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/features/contextual-workflow-reminders.doc.md

Demonstrates how user inputs trigger specific workflow reminders based on pattern matching. The system displays relevant guidance before task execution.

```bash
# User input: "Make a plan for user authentication"
# Triggers: PLANNING_PATTERNS match
# Result: PLANNING_PROMPT displayed with comprehensive planning guidance

# User input: "Investigate how payments work"
# Triggers: INVESTIGATION_PATTERNS match
# Result: INVESTIGATION_PROMPT displayed with discovery workflow guidance
```

--------------------------------

### Gemini API Endpoint Example

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/gemini-video-analysis.md

Example of a POST request to the Gemini API for content generation, specifying the model name.

```bash
POST https://generativelanguage.googleapis.com/v1beta/models/{model-name}:generateContent
```

--------------------------------

### Skill Metadata Example

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills-management.md

Illustrates the YAML frontmatter format used in SKILL.md for defining skill name and description.

```yaml
---
name: Skill Display Name
description: What this skill does and when to use it
---
```

--------------------------------

### Text-to-Video Input Parameters

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/replicate.md

Example of input parameters for a text-to-video generation task, including prompt, duration, resolution, and FPS.

```json
{
  "prompt": "A majestic eagle soaring over mountains",
  "duration": 5,
  "resolution": "720p",
  "fps": 24,
  "seed": 42,
  "guidance_scale": 7.5
}
```

--------------------------------

### Install Required Python Packages

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/skills.md

Install necessary Python packages for PDF processing and other utilities. Ensure your environment is set up before running skills.

```bash
pip install pypdf pdfplumber
```

--------------------------------

### Get Application Limits Endpoint

Source: https://github.com/captaincrouton89/.claude/blob/main/hooks/pushover.md

Retrieve message limits in the response body by making a GET request to the limits endpoint with your application token.

```http
https://api.pushover.net/1/apps/limits.json?token=(your app token)
```

--------------------------------

### Run Google Veo-3 Model

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/replicate-video-models.md

This example demonstrates how to run the Google Veo-3 model for video generation using the Replicate Python client. It specifies a prompt, duration, and resolution for the desired video.

```APIDOC
## Run Google Veo-3 Model

### Description
This example demonstrates how to run the Google Veo-3 model for video generation using the Replicate Python client. It specifies a prompt, duration, and resolution for the desired video.

### Method
`replicate.run()`

### Parameters
#### Model
- **model_name** (string) - Required - The identifier for the model to run (e.g., "google/veo-3").

#### Input Parameters
- **prompt** (string) - Required - The text prompt describing the desired video content.
- **duration** (integer) - Optional - The desired duration of the video in seconds.
- **resolution** (string) - Optional - The desired resolution of the video (e.g., "720p", "1080p").

### Request Example
```python
import replicate

output = replicate.run(
    "google/veo-3",
    input={
        "prompt": "A serene mountain landscape at sunset",
        "duration": 5,
        "resolution": "720p"
    }
)

print(output)
```

### Response
#### Success Response
- The response will contain the output generated by the model, typically a URL to the video file.
```

--------------------------------

### Client Initialization

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/convex-client-api.md

Demonstrates how to initialize the Convex client in JavaScript/TypeScript (Reactive and HTTP), Swift, and React.

```APIDOC
## Client Initialization

### JavaScript/TypeScript ConvexClient (Reactive)
```typescript
new ConvexClient(address: string, options?: ConvexClientOptions)
// Immediately initiates WebSocket connection
```

### JavaScript/TypeScript ConvexHttpClient (Stateless)
```typescript
new ConvexHttpClient(
  address: string,
  options?: {
    skipConvexDeploymentUrlCheck?: boolean,
    logger?: boolean | Logger,
    auth?: string
  }
)
```

### Swift ConvexClient
```swift
let convex = ConvexClient(deploymentUrl: "https://<domain>.convex.cloud")
// Single instance for application lifetime
```

### React ConvexReactClient
```typescript
new ConvexReactClient(deploymentUrl: string, options?: {
  expectAuth?: boolean  // Pause queries until user authenticated
})
```
```

--------------------------------

### Create and List R2 Buckets

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/r2-image-hosting.md

Commands to create a new R2 bucket named 'mystica-assets' and list all available buckets. Use 'list' to verify successful creation.

```bash
# Create the bucket
wrangler r2 bucket create mystica-assets

# List buckets to verify
wrangler r2 bucket list
```

--------------------------------

### Commit Message Formatting Example

Source: https://github.com/captaincrouton89/.claude/blob/main/skills/skills-guide/SKILL.md

Example of how to format commit messages with a type, scope, and subject, followed by a body. This is useful for maintaining a consistent commit history.

```markdown
feat(auth): implement JWT-based authentication

Add login endpoint and token validation middleware
```

```markdown
fix(reports): correct date formatting in timezone conversion
```

--------------------------------

### Test Results and Coverage Report Example

Source: https://github.com/captaincrouton89/.claude/blob/main/hooks/state-tracking/TESTING.md

An example of a test results summary and a detailed coverage report, including overall statistics and specific uncovered lines.

```markdown
## Test Results & Coverage

### ✅ Test Execution

**Tests written:** [N] test cases
**Test suites:** [N] files
**Total assertions:** [N]

**Results:**
```
PASS src/auth/login.test.js
  User Authentication
    when logging in with valid credentials
      ✓ should return JWT token (45ms)
      ✓ should set user session (23ms)
    when logging in with invalid credentials
      ✓ should throw authentication error (12ms)
      ✓ should not create session (8ms)

Test Suites: 1 passed, 1 total
Tests:       4 passed, 4 total
Time:        2.451s
```

--- 

### 📊 Coverage Report

**Overall coverage:**
```
File                | % Stmts | % Branch | % Funcs | % Lines | Uncovered Lines
--------------------|---------|----------|---------|---------|----------------
All files           |   92.5  |   85.3   |   95.0  |   90.8  | 
 auth/              |   95.0  |   88.2   |  100.0  |   94.5  | 
  login.js          |   95.0  |   88.2   |  100.0  |   94.5  | 45, 67
 utils/             |   88.0  |   80.0   |   87.5  |   85.0  | 
  validation.js     |   88.0  |   80.0   |   87.5  |   85.0  | 23-25, 89
```

--- 

### 🎯 Coverage Analysis

**Well-covered areas:**
- ✓ Happy paths: 100% coverage
- ✓ Authentication flows: 95% coverage
- ✓ Error handling: 90% coverage

**Missing coverage:**

1. **auth/login.js:45, 67**
   - Uncovered: Rare database connection error scenario
   - Impact: Low (edge case)
   - Recommendation: Add test or document as acceptable gap
   - Priority: Low

2. **utils/validation.js:23-25**
   - Uncovered: Regex fallback for old browser
   - Impact: Medium
   - Recommendation: Add test for legacy browser path
   - Priority: Medium

**Overall assessment:** Coverage meets [80%+] target. Missing areas are low-priority edge cases.
```

--------------------------------

### Write Tool Output Example (File Update)

Source: https://github.com/captaincrouton89/.claude/blob/main/docs/external/hook-examples.md

Output from the Write tool indicating a file update, including a structured patch detailing the changes.

```json
{
  "content": "test",
  "filePath": "~/Code/Klaude/test.txt",
  "structuredPatch": [
    {
      "lines": [
        "-Test",
        "\\ No newline at end of file",
        "+test",
        "\\ No newline at end of file"
      ],
      "newLines": 1,
      "newStart": 1,
      "oldLines": 1,
      "oldStart": 1
    }
  ],
  "type": "update"
}
```

--------------------------------

### Authenticate and List Models

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/elevenlabs.md

Example using curl to authenticate with an API key and retrieve a list of available models. Ensure you replace 'YOUR_API_KEY' with your actual key.

```bash
curl 'https://api.elevenlabs.io/v1/models' \
  -H 'Content-Type: application/json' \
  -H 'xi-api-key: YOUR_API_KEY'
```

--------------------------------

### POST Request Example (URL-encoded)

Source: https://github.com/captaincrouton89/.claude/blob/main/hooks/pushover.md

This example demonstrates how to construct a URL-encoded POST request to the Pushover API to send a message. Ensure you use HTTPS and the POST method.

```http
POST /1/messages.json HTTP/1.1
Host: api.pushover.net
Content-Type: application/x-www-form-urlencoded
Content-Length: 180

token=azGDORePK8gMaC0QOYAMyEEuzJnyUi&user=uQiRzpo4DXghDmr9QzzfQu27cmVRsG&device=droid4&title=Backup+finished+-+SQL1&message=Backup+of+database+%22example%22+finished+in+16+minutes.
```

--------------------------------

### Generate Image with R2-Hosted Reference Images

Source: https://github.com/captaincrouton89/.claude/blob/main/examples/docs/external/r2-image-hosting.md

Example commands to run the `generate-image.ts` script using R2-hosted reference images via their public R2.dev URLs. Demonstrates using a subset or all available reference images.

```bash
# With R2.dev public URLs (current setup)
npx tsx scripts/generate-image.ts --type "Magic Wand" --materials "wood,crystal" \
  -r "https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_0821.png,https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_4317.png"

# With all reference images
npx tsx scripts/generate-image.ts --type "Fire Staff" --materials "wood,ruby" --provider gemini \
  -r "https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_0821.png,https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_2791.png,https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_4317.png,https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_5508.png,https://pub-1f07f440a8204e199f8ad01009c67cf5.r2.dev/image-refs/IMG_9455.png"
```