dotprompt

Genkit DotPrompt Sample

This sample demonstrates how to use DotPrompt files with Genkit Java for building AI applications with complex inputs and outputs.

Features Demonstrated

• DotPrompt Files: Load and use .prompt files with Handlebars templating
• Complex Input Schemas: Handle nested objects, arrays, and optional fields
• Complex Output Schemas: Parse structured JSON responses into Java objects
• Prompt Variants: Use different prompt variations (e.g., recipe.robot.prompt)
• Partials: Include reusable template fragments (e.g., _style.prompt)

Prompt Files

Located in src/main/resources/prompts/:

• recipe.prompt - Generate recipes with structured output
• recipe.robot.prompt - Robot-themed recipe generation
• story.prompt - Story generation with personality options
• travel-planner.prompt - Complex travel itinerary generation
• code-review.prompt - Code analysis with detailed output
• _style.prompt - Partial template for personality styling

Prerequisites

• Java 21+
• Maven 3.6+
• OpenAI API key

Running the Sample

Option 1: Direct Run

# Set your OpenAI API key
export OPENAI_API_KEY=your-api-key-here

# Navigate to the sample directory
cd java/samples/dotprompt

# Run the sample
./run.sh
# Or: mvn compile exec:java

Option 2: With Genkit Dev UI (Recommended)

# Set your OpenAI API key
export OPENAI_API_KEY=your-api-key-here

# Navigate to the sample directory
cd java/samples/dotprompt

# Run with Genkit CLI
genkit start -- ./run.sh

The Dev UI will be available at http://localhost:4000

Example API Calls

Generate a Recipe

curl -X POST http://localhost:8080/chefFlow \
  -H 'Content-Type: application/json' \
  -d '{"food":"pasta carbonara","ingredients":["bacon","eggs","parmesan"]}'

Robot Chef Recipe

curl -X POST http://localhost:8080/robotChefFlow \
  -H 'Content-Type: application/json' \
  -d '{"food":"pizza"}'

Tell a Story

curl -X POST http://localhost:8080/tellStory \
  -H 'Content-Type: application/json' \
  -d '{"subject":"a brave knight","personality":"dramatic","length":"short"}'

Plan a Trip

curl -X POST http://localhost:8080/planTrip \
  -H 'Content-Type: application/json' \
  -d '{
    "destination": "Tokyo",
    "duration": 5,
    "budget": "$3000",
    "interests": ["food", "culture", "technology"],
    "travelStyle": "adventure"
  }'

Code Review

curl -X POST http://localhost:8080/reviewCode \
  -H 'Content-Type: application/json' \
  -d '{
    "code": "function add(a, b) { return a + b; }",
    "language": "javascript",
    "analysisType": "best practices"
  }'

Output Schemas

Recipe

{
  "title": "Pasta Carbonara",
  "ingredients": [
    {"name": "spaghetti", "quantity": "400g"},
    {"name": "bacon", "quantity": "200g"}
  ],
  "steps": ["Step 1...", "Step 2..."],
  "prepTime": "10 minutes",
  "cookTime": "20 minutes",
  "servings": 4
}

Travel Itinerary

{
  "tripName": "Tokyo Adventure",
  "destination": "Tokyo",
  "duration": 5,
  "dailyItinerary": [
    {
      "day": 1,
      "title": "Arrival & Exploration",
      "activities": [
        {
          "time": "9:00 AM",
          "activity": "Visit Senso-ji Temple",
          "location": "Asakusa",
          "estimatedCost": "$10",
          "tips": "Go early to avoid crowds"
        }
      ]
    }
  ],
  "estimatedBudget": {
    "accommodation": "$800",
    "food": "$500",
    "activities": "$400",
    "transportation": "$300",
    "total": "$2000"
  },
  "packingList": ["Comfortable shoes", "Power adapter"],
  "travelTips": ["Get a JR Pass", "Learn basic Japanese phrases"]
}

Development UI

Access the Genkit Development UI at http://localhost:3100 to:

• Browse available flows and prompts
• Test flows interactively
• View execution traces