everruns-openai

OpenAI LLM provider for Everruns agents.

everruns-openai registers OpenAI drivers into a DriverRegistry from everruns-provider so the same Everruns agent loop can run against OpenAI models. It ships the recommended Responses API driver plus a Chat Completions compatibility driver for OpenAI-compatible endpoints, mapping Everruns' provider-neutral messages, tools, and reasoning onto the OpenAI wire format.

Part of the Everruns ecosystem — the durable agentic harness engine for building unstoppable agents. Providers are swappable: see everruns-anthropic for Claude models, or run with no key at all using the built-in LLM simulator in everruns-runtime.

Quick Example: Agent With OpenAI

use everruns_core::{
    CapabilityRegistry, DriverId, DriverRegistry, PlatformDefinition, ResolvedModel,
};
use everruns_runtime::InProcessRuntimeBuilder;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let mut drivers = DriverRegistry::new();
    everruns_openai::register_driver(&mut drivers);

    let platform = PlatformDefinition::new(CapabilityRegistry::new(), drivers);
    let runtime = InProcessRuntimeBuilder::new()
        .platform_definition(platform)
        .default_model(ResolvedModel {
            model: "gpt-5.4-mini".into(),
            provider_type: DriverId::OpenAI,
            api_key: Some(std::env::var("OPENAI_API_KEY")?),
            base_url: None,
            provider_metadata: None,
        })
        .single_session(|s| {
            s.harness("assistant", "You are a helpful assistant.")
                .agent("openai-agent", "Answer clearly and concisely.")
                .session_title("OpenAI example")
        })
        .build()
        .await?;

    let session_id = runtime.default_session_id().expect("single_session id");
    let result = runtime
        .run_text_turn(session_id, "Write one sentence about reliable agents.")
        .await?;

    println!("{}", result.response);
    Ok(())
}

Driver-Only Example

use everruns_openai::OpenAIChatDriver;

let driver = OpenAIChatDriver::new("your-api-key");
assert!(!driver.uses_custom_url());

What It Provides

A Responses API driver (recommended) and a Chat Completions compatibility driver
Registration into the Everruns DriverRegistry via register_driver
base_url override for OpenAI-compatible endpoints
Streaming, tool calls, and reasoning mapped to provider-neutral Everruns types

Documentation

License