[FEATURE] AI Completions & Advanced Prompting

Problem

Developers need advanced AI completion capabilities with custom prompting, model selection, and configuration options for sophisticated AI interactions.

Goal

Provide full completions API with streaming support, model configuration, and advanced prompting capabilities.

Target Personas

• Developer teams
• System administrators (dotCMS)

Demo Expectations

// What can be demonstrated in bi-weekly demo
const completion = await client.ai.completions.create({
  prompt: 'Analyze this content for key themes',
  indexName: 'blogIndex',
  temperature: 0.7,
  model: 'gpt-4',
  stream: false
});

const config = await client.ai.completions.getConfig();
console.log(config.availableModels); // Available AI models
console.log(completion.response); // AI-generated response

User Stories

• As a developer, I want advanced AI capabilities for complex applications, so that I can build sophisticated AI-powered features with fine-grained control
• As a system administrator, I want to configure AI models and parameters, so that I can optimize AI performance for specific use cases and manage resource usage
• As a developer, I want streaming support for completions, so that I can provide real-time AI responses to users for better experience

Acceptance Criteria

• Completions API with full parameter support (temperature, model, etc.)
• Raw prompt functionality for advanced use cases
• Configuration endpoint for available models and settings
• Streaming support for real-time responses
• Integration with content indexing for context-aware completions
• Support for custom field variables and content filtering
• Comprehensive error handling and timeout management

Links

• dotAI Completions API

Start with the linked dotAI Completions API documentation and the client.ai.completions.create and getConfig entry points shown in the demo. Done means supporting the listed completion parameters, raw prompts, model configuration, streaming, content indexing, filtering, and error and timeout handling.