OpenAIopenai.com
OpenAI's API program excels at API design and usability fundamentals, but two areas are holding back partner and agent success: agent understanding and agent usability. Partners' AI agents lack the structured context, machine-readable docs, and runnable examples they need to integrate correctly without guesswork — and the absence of a self-service developer portal forces unnecessary friction early in the partner journey.
API DesignA clean, typed, well-governed API contract agents can reason about4 pass1 warn0 fail100A+
| Signal | Points | Findings | Rationale | |
|---|---|---|---|---|
| pass | Machine-readable, versioned contractvia spec | 25/25 | Versioning declared (info.version="2.3.0") and url-versioning detected. Investigated: spec 100%, docs 100%. | A current OpenAPI version with a declared versioning scheme lets agents reason about the contract. |
| pass | Auth declared & discoverablevia spec | 25/25 | Authentication required (global or on every operation). Investigated: spec 100%, docs 100%. | Agents can only authenticate when auth is declared, scoped, and discoverable. |
| warn | Schema coverage & depthvia sdk | 25/25 | 7686/7729 (99%) public params are fully typed (no any/**kwargs). Analyzer reports not all methods have a derivable input schema. Investigated: sdk 100%, spec 67%. | Typed, complete request/response schemas are what make agent function-calling possible. |
| pass | Security & governance hygienevia spec | 15/15 | No credential-shaped strings detected in spec. Investigated: spec 100%, wellknown 50%. | No leaked secrets, no critical lint violations, no OWASP API Top-10 spec smells, and a published vulnerability-disclosure channel. |
| pass | Example coveragevia docs | 10/10 | Code samples in 5 language(s) across docs (top: shell, javascript, python); 2/19 pages have ≥ 2 languages on the same page. Investigated: docs 100%, sdk 100%, spec 0%. | Examples carry shape semantics schemas under-specify — for humans and agents alike. |
Developer ExperienceThe context both developers and agents need to integrate fast — onboarding, code samples, complete descriptions and worked examples4 pass0 warn1 fail76B
| Signal | Points | Findings | Rationale | |
|---|---|---|---|---|
| pass | Quickstart presentvia docs | 25/25 | Quickstart at https://developers.openai.com/api/docs/quickstart has a runnable sample in the .md variant (https://developers.openai.com/api/docs/quickstart.md). Investigated: docs 100%. | A quickstart is the fastest path from landing page to first successful call. |
| pass | Code samples in docsvia docs | 18/18 | Code samples in 5 language(s) across docs (top: shell, javascript, python); 2/19 pages have ≥ 2 languages on the same page. Investigated: docs 100%. | Multi-language samples shorten time-to-first-call. |
| pass | Description completenessvia spec | 15/15 | 88% description completeness (target 90%+). Investigated: spec 100%, sdk 73%. | Complete descriptions are the context humans and agents need to use endpoints. |
| pass | Changelog publishedvia spec | 13/13 | changelog documented in the docs site at https://developers.openai.com/api/docs/changelog. Investigated: spec 100%, docs 100%, sdk 100%. | A published changelog lets partners track changes without surprise. |
| fail | Self-service developer portalvia docs | 0/29 | No signup page detected across conventional paths (/signup, /sign-up, /register, /get-started, /console/signup, /dashboard/signup, /try, /try-free, /free, /free-trial, /start, /start-free). Investigated: docs 0%. | Self-serve key/account creation is the fast first call for partners, with no sales gate. |
Agent DiscoveryPartners and their agents can find your APIs — llms.txt, registries, crawlable and reachable docs1 pass3 warn0 fail73B
| Signal | Points | Findings | Rationale | |
|---|---|---|---|---|
| pass | Registry & SDK presencevia docs | 28/28 | Indexed on Context7 (macpaw/openai, 587 snippets). Investigated: docs 100%, cli 100%, sdk 58%. | Listing in MCP registries and publishing SDKs puts the API where agents and their tooling look. |
| warn | llms.txt present, valid & comprehensivevia docs | 18.8/30 | No llms.txt directive found in HTML of any of 50 sampled pages. Investigated: docs 63%, sdk 0%. | A valid, comprehensive llms.txt is the machine-readable entry point for agents. |
| warn | Docs reachable, not hard auth-gatedvia docs | 11.3/30 | 8 of 50 sampled pages return 200 for non-existent URLs (soft 404). Investigated: docs 38%. | Agents can only index and fetch docs they can reach — past auth gates and over correct HTTP semantics. |
| warn | Crawlable / AEOvia wellknown | 9/12 | Sitemap present (1 entries) but carries no lastmod values. Investigated: wellknown 75%. | Bots allowed plus a fresh sitemap make docs findable by agent crawlers. |
Agent UnderstandingAgents can correctly interpret your APIs — machine-readable errors, consistent descriptions, structured data, parseable docs2 pass2 warn2 fail74B
| Signal | Points | Findings | Rationale | |
|---|---|---|---|---|
| pass | Machine-readable errors (RFC 9457)via docs | 28/28 | error codes documented in the docs site at https://developers.openai.com/api/docs/guides/error-codes. Investigated: docs 100%, sdk 100%, spec 50%. | RFC 9457 problem details and a documented error-code inventory let agents parse failures without burning tokens. |
| pass | Operation purpose clarityvia spec | 25/25 | 100% of operations have a clear summary + operationId an agent can select on. Investigated: spec 100%. | Agents select the right endpoint from its summary + operationId; clear, named operations make tool-selection reliable — the strongest driver of correct tool choice. |
| warn | Agent-navigable, token-efficient docsvia docs | 14.3/22 | 16 of 39 pages have substantive content differences between markdown and HTML (avg 36% missing). Investigated: docs 65%. | Server-rendered, clean, small-footprint docs are what an agent can cheaply fetch and parse correctly. |
| warn | Description consistency across surfaces | 1.7/7 | Mean pairwise description similarity across 2 surfaces (spec, docs) is 8% (threshold 35% for full credit). | Every surface tells the same story about what the product is. |
| fail | Agent instructions file (AGENTS.md)via wellknown | 0/10 | No AGENTS.md at the site root or /.well-known/. Investigated: wellknown 0%. | An AGENTS.md gives coding agents explicit setup, auth, and usage instructions to interpret and operate the API — beyond llms.txt's link index. |
| fail | Docs structured datavia docs | 0/8 | Neither JSON-LD nor OpenGraph/meta tags detected across 2 assessed pages (2 JS-rendered). Investigated: docs 0%. | Structured data (JSON-LD/schema.org) on docs pages gives agents an unambiguous parse target and is what answer engines cite. Detected on the JS-rendered head (Firecrawl) for a bounded page budget, so JS-injected JSON-LD is now caught; pages we can't render are excluded rather than failed. |
Agent UsabilityAgents have the context to use your APIs reliably, not just find them4 pass0 warn1 fail92A+
| Signal | Points | Findings | Rationale | |
|---|---|---|---|---|
| pass | Idempotency documentedvia sdk | 27/27 | Built-in retry machinery is present (grep-derived). Investigated: sdk 100%, spec 0%, docs 0%. | Documented idempotency lets agents retry safely. |
| pass | Rate-limit signalingvia spec | 22/22 | rate-limit behavior documented in the docs site at https://developers.openai.com/api/docs/guides/batch. Investigated: spec 100%, docs 100%. | Machine-readable rate-limit headers let agents throttle adaptively. |
| pass | Pagination documented & consistentvia spec | 22/22 | Pagination params detected on 65 list endpoint(s). Investigated: spec 100%, docs 100%, sdk 42%. | Consistent, documented pagination lets agents traverse collections. |
| pass | Sandbox separationvia spec | 20/20 | sandbox documented in the docs site at https://developers.openai.com/api/docs/guides/agents/sandboxes. Investigated: spec 100%, docs 100%, sdk 100%. | An isolated environment lets agents exercise destructive operations safely. |
| fail | Runnable collection with test scriptsvia platform | 0/9 | No public Postman workspace discovered for the org. Investigated: platform 0%. | A public, maintained collection with assertions is runnable truth agents validate against. |
Resources Discovered
The public resources we found for OpenAI — the evidence behind the score. All discovered from public sources; nothing here requires access to your systems.
| Agent hints | Context7 (587) |
|---|---|
| APIs analyzed | 1 — OpenAI API |