Skip to content
Agents API
Agents APIDevelopment preview

Overview

Build applications with a managed agent runtime, durable sessions, and tools.

Compatible with the OpenAI Agents APISame resource model, wire protocol, and official OpenAI SDK. Point the SDK at Rebyte's base URL with a Rebyte API key and model, and it works the same way it does against OpenAI. A few capabilities aren't implemented yet — see Current availability.
On this pageCore conceptsFrom input to resultCurrent availabilityConnect with the OpenAI SDK

Rebyte runs the model and tool loop for your application. You configure an agent, send work to a session, and receive progress and results. Add a managed environment when the task needs code execution or files.

Run your first session →

Core concepts

ConceptPurpose
AgentReusable configuration: model, instructions, tools, and generation settings.
EnvironmentOptional compute and filesystem used by a session.
SessionPersistent execution and conversation state, with a resolved copy of the agent configuration.
Events and itemsEvents communicate inputs and progress; items retain messages and execution output.

A Turn tracks one unit of work inside a session. An Artifact is an immutable deliverable produced by a completed turn. A saved Agent can be reused across Sessions; each managed Session owns its own environment.

From input to result

  1. Create a Session with an inline Agent definition or a saved Agent ID.
  2. Provide input at creation or submit it through the Session events endpoint.
  3. Receive text, tool activity, and lifecycle events as the runtime works.
  4. Read the resulting Items, Turn status, and Artifacts.
  5. Continue the Session with another task, or steer its active Turn.

Your application handles its UI and function tools. Rebyte manages execution and the Session's hosted environment. See Architecture.

Your application starts sessions and receives events and output from the Agents API. Rebyte runs the managed agent harness, which sends tool calls to the Sandbox and receives tool results. Rebyte provisions and manages both the runtime and the Sandbox.

Current availability

These docs describe the development preview, verified against the development service. Production availability has not been established for this implementation. Use the development endpoint supplied for your organization; the quickstart shows the local Relay URL.

CapabilityCurrent implementation
Agents, Sessions, Turns, ItemsImplemented
Live event streaming and steeringImplemented; historical events are not replayed
Managed environments, inline files, ArtifactsImplemented
Functions, HTTP and stdio MCP, VaultsImplemented with documented limits
Web searchImplemented; live mode only, no cached search or location targeting
Inline ZIP skills, plus Rebyte's GitHub-sourced skillsImplemented in managed environments
Webhooks, PluginsNot yet supported
Multi-agent, TracingNot yet supported
Automatic context compactionNot yet supported

The navigation follows OpenAI's guide. A hollow dot marks an unsupported capability. Compatibility is defined by the implemented fields and behavior, not by the presence of a page in this guide.

Connect with the OpenAI SDK

Use client.beta.agents with the Rebyte base URL. The official SDK adds OpenAI-Beta: agents=v1; raw HTTP requests must supply it. The verified Node SDK version is 7.15.0.

Rebyte provides the runtime, models, storage, and billing. The protocol value openai_hosted selects a Rebyte-managed environment when sent to Rebyte; it does not run on OpenAI infrastructure.

Protocol referenceOpenAI Agents API ↗