← All agent briefs
agenticcodingopsdocumentedsession-gated

Operational brief

OpenAI Agents API

Public-beta API for building agents on OpenAI, alongside GPT Live and project-key controls.

OpenAI GitHub organization logo

Choose it when

You are building an agent on OpenAI’s Agents API, not buying a packaged runtime.

Skip it when

You need a finished product rather than an API to build on.

01 / Classification

Autonomy evidence

This is our classification, not a vendor badge. The record says what the product normally does. The gaps stay visible.

How we classify it
Tool use + multi-step plans, still human-gated per session.
Runs unattended
No — the current record treats it as session-gated.
Trigger
Not recorded yet.
State
Not recorded yet.
Tools
OpenAI API
Human control
Not recorded yet.

02 / Operations

Control reality

Useful agents eventually touch something real. This is the operational picture we have, not a promise hidden in a sales page.

Where it runs
cloud
Data boundary
Not recorded yet.
Identity & permissions
Not recorded yet.
What it can write
whatever tools you wire
Surfaces
OpenAI API
Export / lock-in
Not recorded yet.

03 / Receipts

Evidence ledger

Here is what this brief is built on. “Documented” means we found a source. “Tested” means this desk actually ran it. Those are different things.

Evidence status
documented
Source set
1 primary source
Evidence review
Not recorded yet.

Corrections

Not recorded yet. If we change a material claim, the correction belongs here.

Desk notes

What we found

Useful context, not a substitute for the source record above. This is not a desk test unless the evidence status says tested.

Pick this if OpenAI should own the Codex harness, session, and sandbox. Skip it if you wanted the Agents SDK in your app, or the Codex CLI on your machine.

Kind
Hosted Agents API
Header
OpenAI-Beta: agents=v1
GitHub stars
None — not a repo
Product Hunt
No matching listing

What the vendor says

OpenAI’s Agents guide (fetched 2026-09-22) says the Agents API is for long-running tasks where OpenAI manages the agent and saves progress, running a managed Codex harness. The comparison table on that page puts the Agents SDK in your application and the Responses API as direct model calls. The Agents API quickstart says OpenAI manages the agent, conversation, and sandbox; requests need the OpenAI-Beta: agents=v1 header; session keys need api.agents.read, api.agents.write, and api.responses.write. That is a different surface from the open-source Agents SDK (npm @openai/agents / pip openai-agents) and from the Codex CLI.

GitHub

The hosted Agents API is not a GitHub repository, so it has no star count. Do not borrow stars from openai/codex or openai/openai-agents.

Product Hunt

No Product Hunt product page for the OpenAI Agents API turned up in this sweep. Nearby Codex and Operator posts are different products.

Reddit

Anonymous search for “OpenAI Agents API” on 2026-09-22 did not return a thread about this beta. Top hits were unrelated (local inference, Operator from Jan 2025, other “agent API” projects). No community verdict is claimed.

Sources fetched 2026-09-22

04 / Change log

Related coverage

The brief tells you what the product is. Payloads track the changes that may alter the decision.