A production-ready AI research agent built with Rust and the Rig framework. This project is designed for a YouTube tutorial teaching beginners how to build their first AI agent.

Features

Local LLM Support - Uses Ollama for privacy-friendly, free AI inference
Web Search - DuckDuckGo integration (no API key required!)
Tool-Using Agent - Demonstrates agentic AI patterns
Beginner Friendly - Extensive comments explaining Rust patterns
Production Ready - Proper error handling, logging, and CLI

Quick Start

Prerequisites

Install Rust (if not already installed):

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

Install Ollama:
- Visit ollama.ai and follow installation instructions
- Or on Linux: curl -fsSL https://ollama.com/install.sh | sh
Pull a model:
```
ollama pull devstral-small-2
```
Start Ollama:
```
ollama serve
```

Installation

# Copy environment template
cp .env.example .env

# Build the project
cargo build --release

Usage

# Basic research query
cargo run -- "What are the latest developments in Rust async runtime?"

# Quick search mode (no AI synthesis)
cargo run --release -- --quick "Rust web frameworks 2024"

# Use a specific model
cargo run -- --model deepseek-v3.2 "Machine learning in Rust"

# Verbose output
cargo run -- --verbose "WebAssembly trends"

# Show help
cargo run -- --help

Project Structure

rust-research-agent/
├── Cargo.toml          # Project dependencies and metadata
├── .env.example        # Environment variable template
├── README.md           # This file
└── src/
    ├── main.rs         # CLI entry point and application logic
    ├── config.rs       # Configuration management
    ├── agent.rs        # Research agent implementation
    └── tools.rs        # Web search tool (DuckDuckGo)

Configuration

Edit .env to customize the agent:

# Model to use (must be installed in Ollama)
OLLAMA_MODEL=llama3.2

# Ollama server URL
OLLAMA_HOST=http://localhost:11434

# Response creativity (0.0 = focused, 1.0 = creative)
TEMPERATURE=0.7

# Number of search results to analyze
MAX_SEARCH_RESULTS=5

# Logging level
RUST_LOG=info

Learning Rust Concepts

This codebase demonstrates these Rust concepts with inline comments:

Concept	File	Description
Structs & Enums	`config.rs`	Data types and pattern matching
Traits	`tools.rs`	Implementing the Rig `Tool` trait
Ownership & Borrowing	`agent.rs`	Memory safety without GC
Async/Await	`agent.rs`, `tools.rs`	Non-blocking I/O
Error Handling	All files	`Result`, `?` operator, `anyhow`
Derive Macros	All files	`Debug`, `Clone`, `Serialize`
Unit Tests	All files	The `#[cfg(test)]` pattern

Extending the Agent

Adding a New Tool

Create a new struct in tools.rs:
```
pub struct MyNewTool {
    // fields
}
```

Implement the Tool trait:

impl Tool for MyNewTool {
    const NAME: &'static str = "my_tool";
    // ... implement required methods
}

let agent = client
    .agent(&model)
    .tool(web_search_tool)
    .tool(my_new_tool)  // Add here
    .build();

Then set OLLAMA_MODEL in .env or use --model flag.

Testing

# Run all tests
cargo test

# Run with output
cargo test -- --nocapture

# Run specific test
cargo test test_config

Acknowledgments

Rig Framework - The Rust AI framework
Ollama - Local LLM runner
DuckDuckGo - Privacy-respecting search

README.md

Rust - Research Agent