360 check, 9 Nov 2026
Amazon on the pay layer: the 360 agent-readiness check
In short
Amazon scores 40 of 100 on Ora's agent-readiness scan, a D, read 14 Nov 2026. The six payment checks did not apply to amazon.com on this read, which is how Ora treats a site it does not read as a shop. At the door, can't verify agent access. A robots.txt that turns the agents away is the whole pay layer.
On 14 Nov 2026 the public site at amazon.com scored 40 of 100 on Ora's agent-readiness scan, a D, in Ora's E-commerce category. Amazon is the largest online marketplace in the West, with its own agent, Rufus, and a Buy for Me program, so this check reads the Payments layer first: the six checks that look for x402, MPP, UCP, ACP and AP2 on the domain, the same protocols this site's matrix tracks row by row. The other layers follow, with every check Ora ran and what it saw. Ora's one-line summary: Turns agents away at the door: robots.txt blocks every search and user-triggered agent, and there is no MCP server or OpenAPI spec.
The score
| Layer | Points | Share | What the layer asks |
|---|---|---|---|
| Discovery | 10 of 12 | 83% | Can agents find and recommend it |
| Access | 14.67 of 40 | 37% | Can agents read it |
| Usability | 14 of 69 | 20% | Can agents use it |
| Payments | 0 of 0 | none apply | Can agents pay |
| Total | 40 of 100 | 40% | Grade D |
The pay layer
Ora did not run the six payment checks against amazon.com on this read: it marks them not applicable on a site it does not read as a shop, so the layer has no maximum and costs nothing. A marketplace that sells online can still publish any of the five protocols; the matrix says which processors and platforms carry each one. Each protocol links to its matrix row, where the processors and platforms that carry it are dated and sourced.
| Payment check | Result | Points | Matrix row | What Ora saw |
|---|---|---|---|---|
| x402 | Does not apply | 0 of 2 | x402 in the matrix | No x402 payment protocol support detected. No commerce signals detected - agent-payment protocols are optional for non-commerce sites |
| MPP | Does not apply | 0 of 2 | MPP in the matrix | No MPP (Machine Payments Protocol) support detected. No commerce signals detected - agent-payment protocols are optional for non-commerce sites |
| UCP | Does not apply | 0 of 3 | UCP in the matrix | no UCP (Universal Commerce Protocol) signals detected. No commerce signals detected - agent-payment protocols are optional for non-commerce sites |
| ACP | Does not apply | 0 of 3 | ACP in the matrix | no ACP (Agentic Commerce Protocol) signals detected. No commerce signals detected - agent-payment protocols are optional for non-commerce sites |
The other layers, check by check
Every check Ora ran on amazon.com, grouped by layer, with the result, the points and what Ora recorded where its wording passes this site's gate.
Discovery: 10 of 12
| Check | Result | Points | What Ora saw |
|---|---|---|---|
| ARD discovery | Fail | 0 of 1 | No /.well-known/ard.json, no same-site rel="ard" link, and no legacy /.well-known/ai-catalog.json |
| Developer resource discoverability | Pass | 4 of 4 | Pass, no detail recorded. |
| Brand name discoverability | Pass | 3 of 3 | amazon.com ranks #1 for "amazon Books" and #1 for "amazon Magazine" |
| robots.txt AI crawler policy | Fail | 0 of 1 | AI search crawlers: allowed in 0 of 3 checks (3 crawlers on 1 page). AI training crawlers: allowed in 0 of 2 checks (2 crawlers on 1 page), not scored. robots.txt restricts 3/3 evaluated search page/crawler pairs. |
| NPM/PyPI SDK package | Pass | 1 of 1 | Pass, no detail recorded. |
| Agent platform configs | Fail | 0 of 1 | Could not locate a public repo containing agent configs or rules. If our probes missed it, AI coding agents likely will too - link the repo from your homepage or docs so it's discoverable. |
| Agent Plugins manifest | Fail | 0 of 1 | No Agent Plugins manifest (plugin.json) found in discovered repos |
| ChatGPT app listed | Pass | 2 of 2 | Pass, no detail recorded. |
Access: 14.67 of 40
| Check | Result | Points | What Ora saw |
|---|---|---|---|
| Sitemap exists | Fail | 0 of 1 | No sitemap found |
| Content without JavaScript | Partial | 1.67 of 5 | Homepage: 1990 chars of readable text, but no H1 tag. 2 of 2 sampled inner pages have text without JavaScript (3 skipped). |
| Agent discovery file | Fail | 0 of 2 | No agent discovery file found |
| A2A / agent-card | Fail | 0 of 2 | No A2A agent card found at /.well-known/agent-card.json |
| Listed on skills.sh | Fail | 0 of 1 | No official skills self-published on skills.sh (100 community skills with potential relation to this domain found, but only self-published official skills count) |
| pricing.md exists | Fail | 0 of 2 | No pricing.md found - agents need machine-readable pricing to compare products and make purchase recommendations |
| NLWeb Schema Feeds | Fail | 0 of 1 | No schemamap: directive in robots.txt |
| MCP well-known discovery | Fail | 0 of 2 | No MCP server URL discovered |
| HTTP Link headers (RFC 8288) | Fail | 0 of 1 | No HTTP Link response headers (RFC 8288) found |
| Markdown URL fallback | Fail | 0 of 2 | Homepage does not support the .md suffix. Probed https://amazon.com/index.md and it did not return markdown (expected text/markdown content-type or a heading-led non-HTML body). |
| Modular llms.txt per product area | Fail | 0 of 1 | No section-level llms.txt found. Tried /docs/llms.txt, /api/llms.txt, /developers/llms.txt, /guides/llms.txt, /reference/llms.txt, /prime/llms.txt, /deals/llms.txt, /coupons/llms.txt, /blackfriday/llms.txt. |
| robots.txt agent-user policy | Fail | 0 of 1 | robots.txt restricts 3/3 evaluated page/crawler pairs. Checked 3 user-triggered agents (chatgpt-user, claude-user, perplexity-user) on 1 public page. Restricted: chatgpt-user at https://amazon.com/ (Disallow: / |
| llms.txt exists | Fail | 0 of 1 | No llms.txt found at https://amazon.com/llms.txt. Publish it at the site root, where agents that follow the llms.txt spec look for it. |
| llms.txt formatting | Fail | 0 of 2 | No llms.txt found to evaluate quality |
| JSON-LD structured data | Fail | 0 of 2 | No JSON-LD structured data found on homepage |
| Pricing info accessible | Fail | 0 of 3 | No pricing page or pricing data found |
| Public API/docs linked from homepage | Pass | 4 of 4 | Documentation site found at https://docs.amazon.com |
| Agent instruction / when-to-use | Fail | 0 of 1 | No agent instruction file with when-to-use guidance found |
| Skills.sh skill quality | Fail | 0 of 2 | No official skills on skills.sh to evaluate |
| JSON-LD entity linking (sameAs) | Fail | 0 of 1 | No JSON-LD found - no entity linking possible |
| Metadata completeness | Partial | 0.5 of 1 | 3/4 metadata signals present - missing: og:type |
| Organization schema completeness | Fail | 0 of 1 | No JSON-LD found - Organization schema missing |
| Trust anchor pages | Partial | 0.5 of 1 | Only About page verified - missing: Contact, Privacy |
| Markdown alternate link | Fail | 0 of 1 | No sampled page advertises a markdown alternate. Add <link rel="alternate" type="text/markdown" href="..."> (or the equivalent Link response header) pointing at a markdown twin of the page. |
| Redirect hygiene | Pass | 1 of 1 | Pass, no detail recorded. |
| Page token budget | Pass | 1 of 1 | All 3 measured pages fit an agent context budget (largest ~4K tokens) |
| Content starts near the top | Pass | 1 of 1 | Pass, no detail recorded. |
| Content behind auth | Pass | 1 of 1 | All 5 sampled pages are publicly readable (2 with substantive content) |
| OpenAPI spec published | Fail | 0 of 8 | No OpenAPI/Swagger specification found |
| Developer portal | Pass | 5 of 5 | Developer portal found at /developer |
| API catalog (RFC 9727) | Fail | 0 of 2 | No /.well-known/api-catalog found (RFC 9727) |
| Markdown agent docs | Fail | 0 of 1 | Missing markdown content negotiation. Homepage returned text/html when probed with Accept: text/markdown (expected text/markdown), and none of the well-known root paths (/llms.md, /auth.md, /agents.md, /developers.md |
| Markdown content negotiation (acceptmarkdown.com) | Fail | 0 of 1 | Homepage https://amazon.com does not meet the Markdown negotiation requirements: Accept: text/markdown returned text/html;charset=utf-8; Vary header missing Accept (got "user-agent,accept-encoding") |
| Bot-UA markdown serving | Fail | 0 of 1 | No probed bot User-Agent (ChatGPT-User, Claude-User, Perplexity-User) receives markdown when requesting HTML. Optional: detect AI-bot UAs server-side and serve them a markdown representation directly. |
Usability: 14 of 69
| Check | Result | Points | What Ora saw |
|---|---|---|---|
| Public API with reachable endpoints | Pass | 8 of 8 | REST API documentation found at /api. GraphQL endpoint at https://api.amazon.com/graphql is reachable; introspection requires authentication (API key / OAuth), which is expected for agent access. |
| OAuth 2.0 support | Partial | 2 of 5 | OAuth mentioned on homepage but no standard endpoints found |
| Scoped permissions | Fail | 0 of 5 | No declared OAuth scopes, security schemes, or scoped-permission documentation found |
| Agent onboarding friction | Partial | 1 of 2 | Onboarding signals described but not verified live: sandbox/test environment |
| Web Bot Auth directory | Fail | 0 of 2 | No /.well-known/http-message-signatures-directory found |
| OAuth Protected Resource metadata (RFC 9728) | Fail | 0 of 2 | No /.well-known/oauth-protected-resource found (RFC 9728) |
| auth.md exists | Fail | 0 of 2 | No /auth.md found at site root |
| auth.md structure | Fail | 0 of 2 | No /auth.md found at site root |
| Agent auth discovery metadata | Fail | 0 of 3 | Neither /.well-known/oauth-protected-resource nor /.well-known/oauth-authorization-server found. Probed: amazon.com, api.amazon.com. |
| Agent auth WWW-Authenticate hint | Fail | 0 of 1 | No 401 with WWW-Authenticate: Bearer resource_metadata=... found. Probed [https://amazon.com/api, https://amazon.com/api/v1, https://amazon.com/v1, https://amazon.com/v2, https://amazon.com/agent/identity, https://amazon.com/agent/auth |
| agent_auth endpoints reachable | Fail | 0 of 2 | No agent_auth endpoint URIs found in AS metadata or /auth.md prose; nothing to probe |
| MCP server / manifest | Fail | 0 of 10 | No MCP server or manifest found |
| WebMCP support | Fail | 0 of 8 | No WebMCP support detected - no document.modelContext / navigator.modelContext usage found |
| Rate limit response headers | Fail | 0 of 2 | GraphQL introspection auth-gated and no header convention documented on GraphQL-tagged pages |
| Idempotency-Key support | Fail | 0 of 3 | GraphQL introspection auth-gated and no idempotency documentation found on GraphQL-tagged pages |
| JSON error responses | Fail | 0 of 3 | API does not return JSON error responses (or no API detected) |
| NLWeb /ask endpoint | Fail | 0 of 1 | No NLWeb-conformant /ask endpoint found via GET or POST |
| NLWeb streaming support | Fail | 0 of 1 | NLWeb /ask does not support SSE streaming via GET or POST |
| MCP server-card.json | Fail | 0 of 2 | No MCP server card found (checked /.well-known/mcp/server-card.json, the site's ai-catalog.json, servers[].card in mcp.json, and <server>/server-card) |
| Product + docs MCP coverage | Fail | 0 of 2 | This bonus needs two MCP servers: one for your product so agents can take actions, and a separate one for your documentation so agents can answer questions from your docs. |
| Sandbox / test environment | Pass | 2 of 2 | Pass, no detail recorded. |
| GraphQL typed error model | Fail | 0 of 3 | GraphQL introspection auth-gated and no typed graphql error model documentation found on GraphQL-tagged pages |
| GraphQL versioning / deprecation policy | Fail | 0 of 2 | GraphQL introspection auth-gated and no graphql versioning / deprecation policy documentation found on GraphQL-tagged pages |
| GraphQL pagination pattern | Fail | 0 of 2 | GraphQL introspection auth-gated and no graphql pagination pattern documentation found on GraphQL-tagged pages |
| GraphQL async-job pattern | Fail | 0 of 2 | GraphQL introspection auth-gated and no graphql async-job pattern documentation found on GraphQL-tagged pages |
| Agent-friendly 404s | Partial | 1 of 2 | Partial, no detail recorded. |
| Native interactive controls | Pass | 3 of 3 | Pass, no detail recorded. |
| Accessible names on controls | Pass | 2 of 2 | Pass, no detail recorded. |
| Form control labeling | Pass | 2 of 2 | Pass, no detail recorded. |
| Accessibility-tree injection safety (bonus) | Pass | 2 of 2 | Pass, no detail recorded. |
| MCP Apps support | Fail | 0 of 4 | No MCP Apps support detected - this product has an in-agent UI surface but exposes no ui:// resources or tool UI metadata for agents to render. |
| A2UI / generative UI support | Fail | 0 of 2 | No A2UI or generative UI support detected |
| API schema complexity analysis | Fail | 0 of 2 | GraphQL: introspection auth-gated and no schema evidence in docs |
| Function calling compatibility | Fail | 0 of 2 | No API spec found - function calling requires discoverable endpoints |
The fixes worth most points
Ora attaches an estimate to each failing or partial check: how many points the site would gain by passing it. These are the 5 with the largest estimates for Amazon.
- OpenAPI spec published
- Fail, 0 of 8, about 6.7 points to gain. Ora saw:
No OpenAPI/Swagger specification found
. - MCP server / manifest
- Fail, 0 of 10, about 6.4 points to gain. Ora saw:
No MCP server or manifest found
. - WebMCP support
- Fail, 0 of 8, about 5.2 points to gain. Ora saw:
No WebMCP support detected - no document.modelContext / navigator.modelContext usage found
. - Scoped permissions
- Fail, 0 of 5, about 3.2 points to gain. Ora saw:
No declared OAuth scopes, security schemes, or scoped-permission documentation found
. - Content without JavaScript
- Partial, 1.67 of 5, about 2.8 points to gain. Ora saw:
Homepage: 1990 chars of readable text, but no H1 tag. 2 of 2 sampled inner pages have text without JavaScript (3 skipped)
.
Our read
Amazon's robots.txt is the pay layer here. It disallows the OpenAI, Claude and Perplexity search and user agents on every path, which is why the matrix records ACP and UCP on Amazon as not supported, and Ora's own probe was stopped before it could test the door. The six payment checks were run and found no x402, MPP, UCP, ACP or AP2 signal, so the pay layer reads 0 of 0, and the marketplace that sells more online than any other in the West publishes no protocol for an outside agent to buy through, because its policy is that outside agents do not come in. Behind that, the site is thin for agents: 946 characters of home page without an H1, neither a sitemap nor an llms.txt, and no JSON-LD, with a developer portal and a GraphQL endpoint at api.amazon.com for its partners. Rufus and Buy for Me are the company's answer, inside its own walls. The D is a decision, not an oversight.
Test your own checkout
The scan above is free and needs no account: enter a domain at ora.ai and read the same layers for your own site, with the six payment checks mapped to this matrix on the Ora Payments page. The command below runs the same audit from a terminal and exits with 1 under the score you set, so it can gate a deploy. Why Ora, and its limits.
npx ax@0.10 audit https://example.com --min-score 70All 360 checks, one company a day. Corrections to the corrections page.
Sources
- ora.ai/api/score/amazon.com, Ora's scan (vendor), read 14 Nov 2026
- Ora's scoring rules, version 3 (vendor), read 14 Nov 2026
- Ora check catalog (GET /api/checks), read 14 Nov 2026
- amazon.com, the company's site, read 14 Nov 2026
Last reviewed