Mojentic

A modern LLM integration framework for Rust with full feature parity across Python, Elixir, and TypeScript implementations.

Mojentic provides a clean abstraction over multiple LLM providers with tool support, structured output generation, streaming, and a complete event-driven agent system.

🚀 Features

• 🔌 Multi-Provider Support: OpenAI and Ollama gateways
• 🔒 Type-Safe: Leverages Rust's type system for safe LLM interactions
• ⚡ Async-First: Built on Tokio for efficient async operations
• 🛠️ Tool System: Extensible tool calling with automatic recursive execution
• 📊 Structured Output: Generate type-safe structured data with serde
• 🌊 Streaming: Async streaming with Pin<Box<dyn Stream>>
• 🔍 Tracer System: Complete observability for debugging and monitoring
• 🤖 Agent System: Event-driven multi-agent coordination with ReAct pattern
• 📦 24 Examples: Comprehensive examples demonstrating all features

📦 Installation

Add Mojentic to your Cargo.toml:

[dependencies]
mojentic = "1.0.0"
tokio = { version = "1", features = ["full"] }

🔧 Prerequisites

To use Mojentic with local models, you need Ollama installed and running:

1. Install Ollama from ollama.ai
2. Pull a model: ollama pull qwen3:32b
3. Verify it's running: ollama list

🎯 Quick Start

use mojentic::prelude::*;
use std::sync::Arc;

#[tokio::main]
async fn main() -> Result<()> {
    let gateway = Arc::new(OllamaGateway::new());
    let broker = LlmBroker::new("qwen3:32b", gateway, None);

    let messages = vec![
        LlmMessage::user("What is the capital of France?"),
    ];

    let response = broker.generate(&messages, None, None).await?;
    println!("Response: {}", response);

    Ok(())
}

📚 Documentation

• User Guide: mdBook documentation
• API Reference: cargo doc --no-deps --all-features --open

Build documentation locally:

# Build API docs
cargo doc --no-deps --all-features

# Build the mdBook
mdbook build book
open book/book/index.html

🏗️ Architecture

Mojentic is structured in three layers:

Layer 1: LLM Integration

• LlmBroker - Main interface for LLM interactions
• LlmGateway trait - Abstract interface for LLM providers
• OllamaGateway / OpenAiGateway - Provider implementations
• ChatSession - Conversational session management
• TokenizerGateway - Token counting with tiktoken-rs
• EmbeddingsGateway - Vector embeddings
• Comprehensive tool system with 10+ built-in tools

Layer 2: Tracer System

• TracerSystem - Thread-safe event recording
• EventStore - Event persistence and querying
• Correlation ID tracking with Arc sharing
• LLM call, response, and tool events

Layer 3: Agent System

• AsyncDispatcher - Event routing
• Router - Event-to-agent routing
• BaseLlmAgent / AsyncLlmAgent - LLM-powered agents
• IterativeProblemSolver - Multi-step reasoning
• SimpleRecursiveAgent - Self-recursive processing
• SharedWorkingMemory - Agent context sharing with RwLock
• ReAct pattern implementation

🧪 Examples

cargo run --example simple_llm
cargo run --example structured_output
cargo run --example tool_usage
cargo run --example streaming
cargo run --example chat_session
cargo run --example tracer_demo
cargo run --example async_llm
cargo run --example iterative_solver

🔧 Development

# Build
cargo build

# Test
cargo test

# Format
cargo fmt

# Lint
cargo clippy --all-targets --all-features -- -D warnings

# Security audit
cargo deny check

📄 License

MIT License - see LICENSE

Credits

Mojentic is a Mojility product by Stacey Vetzal.