Semantico.AI 3.7.0.1

Semantico.AI

AI-powered extensions for Semantico providing LLM-based documentation generation, alert creation, and autonomous AI actors.

Overview

Semantico.AI is an optional add-on package for Semantico that adds artificial intelligence capabilities:

• Automatic Documentation Generation: Generate comprehensive database documentation using LLMs
• Natural Language Alerts: Convert plain English descriptions to SQL queries
• AI Actors: Autonomous agents that monitor and manage queries/subscriptions
• Multi-Agent Workflows: Parallel documentation generation using specialized domain agents

Installation

dotnet add package Semantico.AI

Supported LLM Providers

• OpenAI (GPT-4, GPT-4 Turbo)
• Anthropic Claude (Claude 3.5 Sonnet, Claude 3 Opus)
• Azure OpenAI (GPT-4 deployments)
• AWS Bedrock (Claude models)

Configuration

Basic Setup

builder.Services.AddSemantico(builder.Configuration, options =>
{
 options.UsePostgreSql(connectionString, "semantico");
 options.UseAI = true; // Enable AI features
});

LLM Configuration (appsettings.json)

{
 "Semantico": {
 "LLM": {
 "Provider": "OpenAI",
 "ApiKey": "your-api-key-here",
 "Model": "gpt-4o",
 "BaseUrl": "https://api.openai.com/v1"
 }
 }
}

Provider Examples

OpenAI:

{
 "Provider": "OpenAI",
 "Model": "gpt-4o",
 "ApiKey": "sk-..."
}

Anthropic Claude:

{
 "Provider": "Anthropic",
 "Model": "claude-3-5-sonnet-20241022",
 "ApiKey": "sk-ant-..."
}

Azure OpenAI:

{
 "Provider": "AzureOpenAI",
 "Model": "gpt-4",
 "ApiKey": "your-azure-key",
 "BaseUrl": "https://your-resource.openai.azure.com"
}

Features

1. Documentation Generation

Generate comprehensive database documentation automatically:

// Via MediatR
var documentation = await mediator.Send(new GenerateDocumentationCommand
{
 DataSourceId = dataSourceId,
 UserId = userId,
 IncludeSampleData = false
});

// Export to PDF, HTML, or Markdown
await mediator.Send(new ExportDocumentationToPdfCommand
{
 DocumentationId = documentation.Id,
 OutputPath = "docs/database.pdf"
});

2. Multi-Agent Documentation

Use multiple specialized AI agents working in parallel:

var documentation = await mediator.Send(new GenerateMultiAgentDocumentationCommand
{
 DataSourceId = dataSourceId,
 UserId = userId,
 MaxConcurrentAgents = 5,
 Progress = progressReporter
});

3. Natural Language Alerts

Create alerts from plain English:

var alert = await mediator.Send(new GenerateAlertCommand
{
 DataSourceId = dataSourceId,
 Description = "Alert me when daily failed logins exceed 100",
 UserId = userId
});

4. AI Actors

Autonomous agents that monitor and optimize your database:

var actor = await mediator.Send(new CreateAiActorCommand
{
 Name = "Performance Monitor",
 Instructions = "Monitor slow queries and create subscriptions for anything over 5 seconds",
 DataSourceId = dataSourceId,
 RequiresApproval = true
});

Dependencies

This package includes:

• Anthropic - Claude API client
• Azure.AI.OpenAI - Azure OpenAI client
• OpenAI - OpenAI API client
• AWSSDK.BedrockRuntime - AWS Bedrock client
• QuestPDF - PDF generation
• Markdig - Markdown processing
• Microsoft.Extensions.AI - Unified AI abstractions

Important Notes

⚠️ AI features are experimental and may produce incorrect or incomplete results. Always review and validate AI-generated content before use in production.

⚠️ API Costs: LLM API calls incur costs. Monitor your usage and set appropriate rate limits.

⚠️ Required Configuration:

• An EncryptionKey must be configured in appsettings.json for secure storage of API keys
• LLM provider API key must be configured

Documentation

For complete documentation, visit: https://github.com/MiBu/semantico

License

MIT License - see LICENSE file for details

Support

• GitHub Issues: https://github.com/MiBu/semantico/issues
• Documentation: https://github.com/MiBu/semantico/wiki

Related Packages

• Semantico.Core - Core library (required)
• Semantico.Core.PostgreSql - PostgreSQL support
• Semantico.Core.SqlServer - SQL Server support
• Semantico.UI - Blazor admin UI