rig

What it is

Rig is a Rust library for building scalable, modular, and ergonomic LLM-powered applications.

How it works

Rig separates portable provider/backend contracts from agent orchestration. The root rig facade re-exports both rig-core (provider-neutral messages, completion models, portable tools, memory and vector-store contracts) and rig-agent (builder, prompt/streaming traits, hooks, tools, and the AgentRun state machine). It supports multiple integrations via per-feature crates (e.g., rig-bedrock, rig-s3vectors, rig-fastembed, rig-qdrant, etc.).

Getting started

cargo add rig
# or: cargo add rig-core

Simple example

use rig::prelude::*;
use rig::providers::openai;

#[tokio::main]
async fn main() -> Result<(), anyhow::Error> {
    // Create OpenAI client
    let client = openai::Client::from_env()?;

    // Create agent with a single context prompt
    let comedian_agent = client
        .agent(openai::GPT_5_2)
        .preamble("You are a comedian here to entertain the user using humour and jokes.")
        .build();

    // Prompt the agent and print the response
    let response = comedian_agent.prompt("Entertain me!").await?;

    println!("{response}");

    Ok(())
}

Note using #[tokio::main] requires you enable tokio's macros and rt-multi-thread features or just full to enable all features (cargo add tokio --features macros,rt-multi-thread).

You can find more examples in each crate's examples directory, and provider-specific integration coverage under tests/providers. See tests/README.md for test target, replay, record, and cassette safety commands. More detailed use case walkthroughs are regularly published on the Rig Dev.to Blog and in Rig's official documentation at rig.rs/docs.

Supported Integrations

The root rig facade exposes companion crates behind one feature per integration. Examples of integration crates include:

• rig-bedrock
• rig-s3vectors
• rig-candle
• rig-vectorize
• rig-fastembed
• rig-gemini-grpc
• rig-vertexai
• rig-helixdb
• rig-lancedb
• rig-memory
• rig-milvus
• rig-mongodb
• rig-neo4j
• rig-postgres
• rig-qdrant
• rig-scylladb
• rig-sqlite
• rig-surrealdb

rig::memory is available without the memory feature; it contains core memory traits and an in-memory backend re-exported from rig-core. Enabling features = ["memory"] adds reusable history-shaping policy types from the rig-memory crate.

We also have rig-onchain-kit for Rig Onchain Kit functionality.

Get started (additional context)

Rig targets both portable provider contracts and agent orchestration, with explicit separation between rig-core (providers, memory, vector stores, etc.) and rig-agent (agent runtime, prompts, hooks, and state machine).

Features (highlights)

• Agentic workflows with multi-turn streaming and prompting
• Default classic agent runtime
• GenAI Semantic Convention compatibility
• 20+ model providers under one interface
• 10+ vector store integrations under one interface
• Full support for LLM completion and embedding workflows
• Support for transcription, audio generation and image generation model capabilities
• WASM support for portable core and classic runtime (target matrix detailed in crate READMEs)