Crate ggen_ai 

ggen-ai - LLM integration layer for ggen

Thin wrapper around genai for ggen with environment support and caching.

This crate provides a simplified LLM integration layer for ggen, focusing on:

• Environment-based configuration
• Response caching
• Template generation
• SPARQL query generation
• Ontology generation
• Code refactoring assistance

Features

• Multi-provider LLM support: OpenAI, Anthropic, Ollama, Gemini, DeepSeek, xAI/Grok, Groq, Cohere (via genai)
• Environment-based configuration: Automatic API key detection and model selection
• Response caching: Reduce API costs and latency with intelligent caching
• Template generation: Natural language to ggen templates
• SPARQL query generation: Intent-based query construction
• Ontology generation: Domain descriptions to RDF/OWL
• Code refactoring: AI-assisted code improvement suggestions
• RDF-based CLI generation: Generate CLI projects from RDF ontologies

Quick Start

use ggen_ai::{GenAiClient, LlmClient, LlmConfig};

// Initialize client with default configuration
let config = LlmConfig::default();
let client = GenAiClient::new(config)?;

// Generate response
let response = client.complete("Explain Rust ownership").await?;
println!("{}", response.content);

Module Organization

• cache - LLM response caching
• client - LLM client abstraction
• config - Configuration management
• generators - Specialized generators (templates, SPARQL, ontologies)
• providers - LLM provider implementations
• prompts - Prompt templates and builders
• rdf - RDF-based CLI generation
• security - API key masking and security
• streaming - Streaming response support
• types - Type definitions

Re-exports

pub use cache::CacheConfig;

pub use cache::CacheStats;

pub use cache::LlmCache;

pub use client::GenAiClient;

pub use client::LlmChunk;

pub use client::LlmClient;

pub use client::LlmConfig;

pub use client::LlmResponse;

pub use client::UsageStats;

pub use config::get_global_config;

pub use config::init_global_config;

pub use config::AiConfig;

pub use config::GlobalLlmConfig;

pub use config::LlmProvider;

pub use error::GgenAiError;

pub use error::Result;

pub use generators::NaturalSearchGenerator;

pub use generators::OntologyGenerator;

pub use generators::QualityMetrics;

pub use generators::RefactorAssistant;

pub use generators::SparqlGenerator;

pub use generators::TemplateGenerator;

pub use generators::TemplateValidator;

pub use generators::ValidationIssue;

pub use providers::adapter::ollama_default_config;

pub use providers::adapter::ollama_ministral_3b_config;

pub use providers::adapter::ollama_qwen3_coder_config;

pub use providers::adapter::MockClient;

pub use rdf::Argument;

pub use rdf::ArgumentType;

pub use rdf::CliGenerator;

pub use rdf::CliProject;

pub use rdf::Dependency;

pub use rdf::Noun;

pub use rdf::QueryExecutor;

pub use rdf::RdfParser;

pub use rdf::TemplateRenderer;

pub use rdf::Validation;

pub use rdf::Verb;

pub use security::MaskApiKey;

pub use security::SecretString;

pub use streaming::StreamConfig;

pub use types::DecisionId;

pub use types::PolicyId;

pub use types::RequestId;

pub use types::RuleId;

Modules

cache

LLM response caching with Moka

client

Simplified client interface using rust-genai

config

Configuration management for ggen-ai

constants

Constants for ggen-ai

error

Error types for ggen-ai

error_utils

Error handling utilities for ggen-ai

generators

AI-powered generators for ggen

parsing_utils

Code block parsing utilities for ggen-ai

prompts

Prompt engineering for AI-powered generators

providers

LLM provider implementations

rdf

RDF-based CLI project generation.

security

Security utilities for protecting sensitive data

streaming

LLM Streaming Support via rust-genai

types

Type-safe wrappers for ggen-ai

Constants

VERSION

Version information

Functions

init_logging

Initialize tracing for the ggen-ai crate