Simple LLM Client

A Rust crate for interacting with Large Language Model APIs to streamline content creation, research, and information synthesis for use with RAG applications.

Features

• Perplexity AI Integration: Seamlessly connect with the Perplexity AI API for advanced research capabilities
• Markdown Output: Automatically format responses as Markdown with proper citation formatting
• Streaming Support: Option to stream responses in real-time or receive complete responses
• Citation Handling: Extract and format citations from AI responses
• Multiple Provider Support: Future updates will include additional AI providers (OpenAI, Anthropic, Google, etc.)

Installation

Add this to your Cargo.toml:

[dependencies]
simple-llm-client = "^0.2"

The crate is also available via its directory name for local development:

[dependencies]
simple_llm_client = { path = "path/to/llm_client" }

Usage

Basic Example

use simple_llm_client::perplexity::{chat_completion, models::ChatMessage};
use std::error::Error;

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
    let messages = vec![
        ChatMessage {
            role: "system".to_string(),
            content: "Be precise and concise.".to_string(),
        },
        ChatMessage {
            role: "user".to_string(),
            content: "How many stars are there in our galaxy?".to_string(),
        },
    ];

    // Stream the response to stdout
    chat_completion("sonar-pro", messages).await?;
    Ok(())
}

Markdown Output Example

use simple_llm_client::perplexity::{chat_completion_markdown, models::ChatMessage};
use std::{error::Error, path::Path};

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
    let messages = vec![
        ChatMessage {
            role: "system".to_string(),
            content: "Be precise and concise. Return the response as markdown.".to_string(),
        },
        ChatMessage {
            role: "user".to_string(),
            content: "Explain the difference between fusion and fission.".to_string(),
        },
    ];

    // Save the formatted response to a file
    chat_completion_markdown("sonar-pro", messages, Some(Path::new("./output")), "research_result.md").await?;
    Ok(())
}

Configuration

The crate requires a Perplexity API key to be set in your environment or in a .env file:

PERPLEXITY_API_KEY=your_api_key_here
OPENAI_API_KEY=your_api_key_here

For the examples to work correctly, create a .env file in the project root with your API key.

Directory Structure

When using the file output functionality, make sure the output directories exist:

mkdir -p output  # Create the output directory if it doesn't exist

The example code creates this directory automatically.

Roadmap

• Additional Providers: Support for other AI research and generation APIs will be added in future releases:
  • Anthropic (Claude models)
  • Google (Gemini models)
  • Others based on community demand
• Advanced Formatting Options: Customizable output formatting and templates
• Citation Style Options: Support for different citation styles (APA, MLA, etc.)
• Context Management: Tools for managing conversation context and history
• Multi-provider Research: Aggregate and compare responses from multiple providers

License

This project is licensed under the MIT License - see the LICENSE file for details.

Examples

The crate includes several examples to help you get started:

Running Examples

To run any example, use the cargo run --example command:

# Test the Perplexity implementation
cargo run --example perplexity --features perplexity

# Test the OpenAI implementation
cargo run --example openai --features openai

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.