← OpenAI

Agent report card

Can an agent use OpenAI API through API service lifecycle?

Grade B: Machine credentials exist but require manual copying, broad authority, or weak lifecycle support.

Observedrev sha256:214cf1b1dc3b…barrier checks 0/5 verified

Agent ReadinessBLimited

Five-stage assessment

Where agent autonomy holds and where it breaks

Five states · each from its own evidence · lifecycle order, not a journey
  1. EvaluateReady
  2. Sign upLimited
  3. PayReady
  4. ProvisionLimited
  5. OperateLimited
  1. 01

    Evaluate

    Ready

    The exact service and stable evaluation or access entrypoints are publicly discoverable.

    The API Overview reference details stable public evaluation and entrypoints, including HTTP Bearer authentication and base endpoint URLs like https://api.openai.com/v1/models.

    Applicable terms are stable, readable, retrievable, and materially complete.

    Eligibility conditions and required inputs are explicit and decidable.

    Price or no-charge status, variables, currency, and material conditions are explicit.

    Verified structured artifacts accelerate discovery of the evaluation surface.

    Evidence and actions6 signals

    Can an agent decide every material eligibility condition before commitment?

    supported: fresh

    Eligibility conditions and required inputs are explicit and decidable.

    The OpenAI Services Agreement explicitly states that it applies to business and developer use of OpenAI's APIs and other business services, requiring that the customer organization is lawfully able to enter into the agreement and that the individual accepting it has legal authority to bind the entity.

    eligibility_decidability: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent estimate cost or confirm no-charge status before commitment?

    supported: fresh

    Price or no-charge status, variables, currency, and material conditions are explicit.

    OpenAI specifies prepaid billing rules including minimum purchase amounts in USD, alongside explicit per-unit rates and tool usage fees.

    pricing_decidability: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent find the exact service and its stable entrypoints?

    supported: fresh

    The exact service and stable evaluation or access entrypoints are publicly discoverable.

    The API Overview reference details stable public evaluation and entrypoints, including HTTP Bearer authentication and base endpoint URLs like https://api.openai.com/v1/models.

    service_discovery: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Are verified structured discovery artifacts available?

    supported: fresh

    Verified structured artifacts accelerate discovery of the evaluation surface.

    The OpenAI API reference explicitly mentions the availability of llms.txt as a structured complete documentation index.

    structured_evaluation_discovery: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent retrieve and understand the applicable commitment terms?

    supported: fresh

    Applicable terms are stable, readable, retrievable, and materially complete.

    The applicable OpenAI Services Agreement is retrievable, human-readable, stable, and sets out complete commitment terms, customer obligations, and usage terms for business and API services.

    terms_access: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Does the vendor deliberately describe access for web agents?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The retained evidence from the API reference and quickstart pages does not contain explicit declarations or rules regarding web crawler/bot access policies (such as robots.txt rules or agent crawling specifications).

    verified_web_agent_access: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

  2. 02

    Sign up

    Limited

    Machine credentials exist but require manual copying, broad authority, or weak lifecycle support.

    OpenAI API keys can be created and deleted via the API key page, but the secret key is displayed only upon creation and requires manual saving and updating in applications.

    A stable canonical route begins the required access bootstrap.

    Evidence and actions5 signals

    Can an agent operate the access controls and safe handoffs deterministically?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The retained evidence for the registration page contains only header title text ('OpenAI Platform') without visible form controls, fields, or handoffs, leaving access control operability unresolved.

    access_control_operability: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Is there a stable route to begin obtaining service access?

    supported: fresh

    A stable canonical route begins the required access bootstrap.

    A stable canonical route to access the platform exists at https://platform.openai.com/signup.

    access_entrypoint_stability: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent obtain access without an unsupported CAPTCHA boundary?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The captured evidence for the signup page displays minimal page header metadata and title text ('OpenAI Platform') without revealing whether CAPTCHA challenges are present or omitted.

    captcha_compatible_access: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent obtain scoped, revocable authority for this service?

    supported: fresh

    Machine credentials exist but require manual copying, broad authority, or weak lifecycle support.

    OpenAI API keys can be created and deleted via the API key page, but the secret key is displayed only upon creation and requires manual saving and updating in applications.

    delegated_identity_access: Limited: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can required phone verification be completed through a supported boundary?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The captured evidence for the platform signup page contains only generic header metadata and title text ('OpenAI Platform'), leaving phone verification requirements unresolved.

    phone_verification_compatible: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    What would improve this stage

    • Provide scoped delegated authorization with consent, revocation, and deterministic resumption.
  3. 03

    Pay

    Ready

    Charge or no-charge status, currency, recurrence, and material conditions are disclosed.

    OpenAI's prepaid billing page discloses the commitment before purchase: a minimum credit purchase of $5, a default of $10, auto-reload settings reviewed before confirming, and credits that expire after one year and are non-refundable.

    A documented direct purchase path reaches paid access without a vendor decision point.

    Evidence and actions4 signals

    Can an agent construct checkout, hand off approval safely, and resume?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The retained evidence describes setting up prepaid billing and adding payment details in the organization's billing overview, but it does not establish a programmatic or agent-operable deterministic checkout, handoff, and resumption flow.

    checkout_operability: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Is the exact commercial commitment disclosed before authorization?

    supported: fresh

    Charge or no-charge status, currency, recurrence, and material conditions are disclosed.

    OpenAI's prepaid billing page discloses the commitment before purchase: a minimum credit purchase of $5, a default of $10, auto-reload settings reviewed before confirming, and credits that expire after one year and are non-refundable.

    commitment_disclosure: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can payment be authorized within scoped agent or explicit human authority?

    supported: fresh

    Current admissible evidence does not resolve this finding.

    The evidence notes that an account manager must have permission to manage billing and complete payment details, but it does not document a delegated payment authority rail or scoped agent authorization mechanism.

    payment_authorization: Unknown: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Is a direct self-service path to paid access documented?

    supported: fresh

    A documented direct purchase path reaches paid access without a vendor decision point.

    Users with billing permission can access their API billing overview, add payment details, select a credit amount, and confirm the purchase directly to reach paid API access without a vendor decision point or sales approval.

    self_service_purchase: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

  4. 04

    Provision

    Limited

    Delivery needs additional human transfer or has weak scope or lifecycle semantics.

    OpenAI requires creating an API key in the dashboard or API key page, where full secret keys are displayed only upon creation and must be saved manually. Because secret delivery relies on manual human dashboard retrieval rather than a programmatic delivery path, access material delivery is partial.

    Provisioning is machine-triggerable or follows deterministically from an allowed handoff.

    A synchronous success response returns a usable created resource or access material; otherwise asynchronous work exposes progress, terminal state, reconciliation, and a documented bound.

    Evidence and actions3 signals

    Can usable access material be delivered securely to an authorized agent?

    supported: fresh

    Delivery needs additional human transfer or has weak scope or lifecycle semantics.

    OpenAI requires creating an API key in the dashboard or API key page, where full secret keys are displayed only upon creation and must be saved manually. Because secret delivery relies on manual human dashboard retrieval rather than a programmatic delivery path, access material delivery is partial.

    access_material_delivery: Limited: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent determine successful provisioning completion and reconcile asynchronous failure?

    supported: fresh

    A synchronous success response returns a usable created resource or access material; otherwise asynchronous work exposes progress, terminal state, reconciliation, and a documented bound.

    A synchronous API call to the Responses API returns a usable created response object containing a terminal status ('completed') and output text payload.

    provisioning_completion: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can provisioning be initiated within supported agent authority?

    supported: fresh

    Provisioning is machine-triggerable or follows deterministically from an allowed handoff.

    API requests can be initiated programmatically within supported agent authority by making POST HTTP requests to /v1/responses or using official OpenAI SDK client creation methods.

    provisioning_operability: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    What would improve this stage

    • Deliver scoped access material through a secure documented machine-usable flow.
  5. 05

    Operate

    Limited

    Safe failure handling is documented only partially.

    OpenAI documents error codes and retry guidance following Retry-After; the selected quotes do not establish idempotency, cancellation or reconciliation semantics for the Responses operation.

    Only part of the credential lifecycle is agent-operable.

    Every essential target has a stable documented agent-usable interface alternative.

    Request-time authentication is documented and usable under scoped agent authority.

    Essential operations have stable readable request, response, and effect semantics.

    A verified agent-native interface covers the declared assessment targets.

    Evidence and actions6 signals

    Is an agent-native protocol interface verified against the essential targets?

    supported: fresh

    A verified agent-native interface covers the declared assessment targets.

    The Responses API provides a unified interface for building agent-like applications with built-in tools such as web search, file search, computer use, code interpreter, and remote MCPs.

    agent_protocol_interface: Ready: tested 6 Sept

    • api-service

    Method public-semantic-assessment 2026-09-06

    Can an agent manage credential expiry, rotation, revocation, compromise, and recovery?

    supported: fresh

    Only part of the credential lifecycle is agent-operable.

    OpenAI allows creating and deleting API keys via the API key dashboard page, but key management relies on manual UI actions.

    credential_lifecycle: Limited: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent handle applicable failure modes safely?

    supported: fresh

    Safe failure handling is documented only partially.

    OpenAI documents error codes and retry guidance following Retry-After; the selected quotes do not establish idempotency, cancellation or reconciliation semantics for the Responses operation.

    failure_contract: Limited: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Can an agent authenticate service operations with scoped authority?

    supported: fresh

    Request-time authentication is documented and usable under scoped agent authority.

    Requests can be authenticated using HTTP Bearer authentication passing an API key via the Authorization header.

    operation_authentication: Ready: tested 6 Sept

    Method public-semantic-assessment 2026-09-06

    Are essential operation inputs, outputs, and effects stable and decidable?

    supported: fresh

    Essential operations have stable readable request, response, and effect semantics.

    The documentation details stable request parameters, response schemas, error structures, and tool call conventions across essential operations.

    operation_contract: Ready: tested 6 Sept

    • api-service

    Method public-semantic-assessment 2026-09-06

    Can an agent perform every essential assessment target through a usable interface?

    supported: fresh

    Every essential target has a stable documented agent-usable interface alternative.

    Usable interfaces including client libraries, HTTP REST endpoints, and the Responses API are documented across all essential assessment targets.

    target_interface_access: Ready: tested 6 Sept

    • api-service

    Method public-semantic-assessment 2026-09-06

    What would improve this stage

    • Document applicable structured errors, retry, cancellation, and reconciliation semantics.
    • Provide scoped expiry, rotation, revocation, compromise, and recovery operations.

What works

  • Evaluate: The exact service and stable evaluation or access entrypoints are publicly discoverable.
  • Pay: Charge or no-charge status, currency, recurrence, and material conditions are disclosed.

What blocks agents

No stage is established as blocked.

Where autonomy is limited

  • Sign up: Machine credentials exist but require manual copying, broad authority, or weak lifecycle support.
  • Provision: Delivery needs additional human transfer or has weak scope or lifecycle semantics.
  • Operate: Safe failure handling is documented only partially.

What this grade establishes

14 of 14 graded signals have current supporting evidence. 0 of 5 additional barrier checks are verified.Unknown checks are not a pass or a confirmed blocker. This report is not proof of an end-to-end purchase or service session.

How the grade is derived

A through C grades count Limited stages; D and F reflect Blocked stages by lifecycle severity.

Coverage is complete only when every graded signal has supported, fresh, non-conflicting evidence. Barrier and informational signals do not change coverage; only fresh, supported barrier values that passed their admission evidence rule participate in stage outcomes.

Each stage takes its worst graded signal or verified barrier. A fresh, supported mandatory barrier that passed its admission evidence rule can block that stage; incomplete or stale barrier evidence remains context only. The overall grade is derived from the five stage states.

Startup offers

OpenAI on the startup-offers board

Distribution

The grade’s badge

One stable image resolving the record’s current live facts, linking to the canonical page. Embed it once; it keeps itself current. Placement is never evidence and never affects rank.

Live sourcey badge for OpenAI Agent Readiness grade B
[![sourcey record](https://sourcey.com/badges/agent-readiness/arp_01kzf0m4xe2b7nt9w4r6dh1q8c.svg)](https://sourcey.com/log/openai/agent-readiness/openai-api/api-service-lifecycle)

Improve the report

Correct it, rerun it, or improve the funnel.

Profile
arp_01kzf0m4xe2b7nt9w4r6dh1q8c
Projection
sha256:90b24fcbda2b2d323316d400b496ebabdfbd7052ea1c47177063a6565b16baf7
Policy
service-use-2026-09-07-blocking-barriers-r12 · sha256:fc9e8f046d86fcf19ab2f1bc547219e503cdc11ffa135d020db7285965cd7d1d