Agent MCP Directory

Vercel: the 360 agent-readiness check

Vercel scores 91 of 100 on Ora's agent-readiness scan, an A, read 9 November 2026. Of the five surfaces an agent acts on, Ora found 4: an OpenAPI specification, a public API, OAuth, a command line. A 91 whose remaining points are scope names and rate-limit headers.

On 9 November 2026 the public site at vercel.com scored 91 of 100 on Ora's agent-readiness scan, an A, in Ora's Infrastructure category. Ora probes a site the way an agent would: it requests pages with the agents' own user agents, reads robots.txt, looks for the files agents read, and tries the API, MCP and payment surfaces. This check reads the act layer first, because that is what the directory records, then the door, what an agent can read, and whether the company is found and recommended. Ora's one-line summary: Vercel offers strong discovery and access with a verified zero-auth API, and WebMCP support plus named OAuth scopes would round out agent readiness.

The score

Vercel on Ora, read 2026-11-09: points per layer, with the share of each layer's maximum. 68 checks pass, 8 are partial, 21 fail and 35 do not apply to this site.
LayerAsksPointsShare
UsabilityCan agents use it73 of 8982%
AccessCan agents read it62 of 6497%
DiscoveryCan agents find and recommend it21 of 21100%
PaymentsCan agents pay0 of 0none apply
TotalGrade A91 of 10091%

What an agent can do

The surfaces an agent acts on: an API it can find and call, an MCP server, an agent card, OAuth metadata, a command line. This is the layer the directory records, so it comes first.

In the directory, Vercel sign-in required, tools not seen, not counted; the connector run of 4 November 2026 is on the entry page. Ora's MCP server check for the same domain reads partial; the two look for different things, the directory for a server that answers a handshake and lists tools, Ora for a server it can find from the site.

What an agent can do: 19 checks Ora ran on vercel.com, read 2026-11-09. The last column is Ora's own observation where it published one.
CheckResultPointsWhat Ora saw
MCP server / manifest mcp-serverPartial5 of 6MCP server detected at https://mcp.vercel.com; it requires authentication. Presence is verified (5/6), but Ora could not inspect its tools or verify access scopes without credentials.
MCP well-known discovery mcp-well-known-discoveryPartial1 of 2MCP server at https://mcp.vercel.com - list its server card, hosted on your own domain, in /.well-known/ai-catalog.json so MCP clients can discover it
Public API with reachable endpoints public-apiPass7 of 7REST API documentation found at https://vercel.com/docs/getting-started-with-vercel.md?from=llms-txt. Best-of-protocols score: 7/7.
Public API/docs linked from homepage public-api-docsPass3 of 3Documentation site found at https://vercel.com
Developer portal developer-portalPass6 of 6Developer portal found at /docs
OpenAPI spec published openapi-specPass7 of 7OpenAPI spec found at https://vercel.com/openapi.json (version: 3.0.3)
API catalog (RFC 9727) api-catalog-rfc9727Pass2 of 2no detail published
A2A / agent-card a2a-agent-cardFail0 of 2No A2A agent card found at /.well-known/agent-card.json
Agent discovery file agent-discovery-filePass2 of 2Agent Skills index (agentskills.io) found at /.well-known/agent-skills/index.json with 9 skill(s)
OAuth 2.0 support oauth-supportPass5 of 5OAuth authorization server metadata at https://vercel.com
OAuth Protected Resource metadata (RFC 9728) oauth-protected-resourcePass2 of 2no detail published
auth.md exists auth-md-existsFail0 of 2Path https://vercel.com/auth.md returned HTML (SPA / 404 fallback), not an auth.md document
WebMCP support webmcpFail0 of 5No WebMCP support detected - no document.modelContext / navigator.modelContext usage found (scanned 43 of 43 same-origin script bundle(s) for modelContext registrations)
NLWeb /ask endpoint nlweb-askFail0 of 1No NLWeb-conformant /ask endpoint found via GET or POST
CLI tool available cli-toolPass3 of 3CLI package found on npm: vercel
Multi-language SDK packages rest-sdk-packagesPartial2 of 3SDK found in npm, additional SDKs mentioned at https://vercel.com
Rate limit response headers rate-limit-headersFail0 of 2No REST rate-limit headers found on probed endpoints
Idempotency-Key support idempotency-key-supportFail0 of 3No REST Idempotency-Key support detected
JSON error responses json-error-responsesPass4 of 4no detail published

At the door

What the site answers when a request arrives with an agent's user agent, and what robots.txt says to agents.

At the door: 8 checks Ora ran on vercel.com, read 2026-11-09. The last column is Ora's own observation where it published one.
CheckResultPointsWhat Ora saw
Not blocked by bot detection bot-detectionPass2 of 2Site accessible to 4 agent user-agents (ChatGPT-User, OAI-SearchBot, Claude-User, PerplexityBot).
Agent crawler reachability agent-crawler-reachabilityPass2 of 2Reachable to every agent we tested - ChatGPT-User: reachable, OAI-SearchBot: reachable, Claude-User: reachable, PerplexityBot: reachable
robots.txt AI crawler policy robots-ai-policy-qualityPass2 of 2AI search crawlers: allowed in 18 of 18 checks (3 crawlers on 6 pages). AI training crawlers: allowed in 12 of 12 checks (2 crawlers on 6 pages), not scored.
robots.txt agent-user policy robots-agent-user-policyPass2 of 2No applicable crawl or usage restriction found for 3 user-triggered agents (chatgpt-user, claude-user, perplexity-user) on 1 public page: https://vercel.com/. Declared policy only;
Content behind auth docs-auth-gatePass2 of 2All 5 sampled pages are publicly readable (5 with substantive content)
Homepage not hidden from search noindex-directivesPass1 of 1No noindex on the homepage: no robots, googlebot or bingbot meta tag and no X-Robots-Tag header hides it from search.
Redirect hygiene redirect-hygienePass1 of 1no detail published
Web Bot Auth directory web-bot-auth-directoryFail0 of 2No /.well-known/http-message-signatures-directory found

What an agent can read

Whether the text is in the first response, what the site publishes for agents, and how it describes itself to them.

What an agent can read: 17 checks Ora ran on vercel.com, read 2026-11-09. The last column is Ora's own observation where it published one.
CheckResultPointsWhat Ora saw
Content without JavaScript content-no-jsPartial2 of 3Homepage: 1009 chars with H1, but 1.0% content ratio is below the 5% target. 4 of 4 sampled inner pages have text without JavaScript (1 skipped).
Page token budget page-token-budgetPass1 of 1All 7 measured pages fit an agent context budget (largest ~4K tokens)
Sitemap exists sitemapPass2 of 2Valid sitemap found at https://vercel.com/sitemap.xml with 9388 entries
Sitemap freshness (lastmod) sitemap-lastmodPass1 of 1no detail published
llms.txt exists llms-txt-existsPass1 of 1Found the llms.txt at https://vercel.com/llms.txt.
llms.txt formatting llms-txt-formattingPass2 of 2The llms.txt is well-formatted: 48 lines with markdown links, 5,397 characters in total.
llms.txt links resolve llms-txt-links-resolvePass2 of 2no detail published
Markdown agent docs markdown-negotiationPass1 of 1Path-suffix markdown docs served with text/markdown content-type: /agents.md
Markdown URL fallback markdown-url-fallbackFail0 of 2Homepage does not support the .md suffix. Probed https://vercel.com/index.md and it did not return markdown (expected text/markdown content-type or a heading-led non-HTML body).
Bot-UA markdown serving agent-ua-markdownFail0 of 1No 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.
Agent mode view agent-mode-viewFail0 of 2?mode=agent returns same content as homepage (no dedicated agent view)
JSON-LD structured data json-ldPass4 of 4Rich JSON-LD identity: Organization with name, description, url, and sameAs/logo/address (3 block(s))
Organization schema completeness org-schema-completenessPass2 of 2no detail published
JSON-LD entity linking (sameAs) json-ld-entity-linkingPass2 of 2no detail published
Schema type breadth schema-type-breadthPartial1 of 2Some extended schema types found: Service - add 2 more types that match what the homepage shows, for full coverage
Metadata completeness metadata-completenessPass2 of 2All metadata signals present: canonical URL, lang="en", og:image, og:type
Agent instruction / when-to-use agent-instructionPass3 of 3no detail published

Found and recommended

Whether an assistant can find the company by name, confirm what it is, and see it in the registries agents read.

Found and recommended: 9 checks Ora ran on vercel.com, read 2026-11-09. The last column is Ora's own observation where it published one.
CheckResultPointsWhat Ora saw
Brand name discoverability brand-search-accuracyPass3 of 3vercel.com ranks #3 for "Vercel Agentic infrastructure platform"
Wikipedia / Wikidata entity presence wikipedia-presencePass4 of 4Wikipedia "Vercel" and Wikidata Q56069184 both verified - domain confirmed on both sources
Developer resource discoverability agentic-search-specificPass3 of 3no detail published
Listed in MCP registries mcp-registry-listedPass1 of 1no detail published
NPM/PyPI SDK package npm-sdk-packagePass1 of 1no detail published
ChatGPT app listed chatgpt-app-listedPass2 of 2no detail published
Registry branding registry-brandingFail0 of 2No registry branding (name, icon, description) found
Listed on skills.sh skills-sh-listedPass1 of 1no detail published
ARD discovery ard-catalogPass1 of 1no detail published

Payments

The agent payment protocols Ora probes for. A site that sells nothing has nothing here, and Ora scores the layer out of zero when none applies. None of the 6 payment checks applied to vercel.com.

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 Vercel, with what Ora observed. The wording of each fix is Ora's; the 28 checks it runs on an MCP server are listed with their layer and tier.

  • WebMCP support: fail, 0 of 5, about 2.5 points to gain. Ora saw: No WebMCP support detected - no document.modelContext / navigator.modelContext usage found (scanned 43 of 43 same-origin script bundle(s) for modelContext registrations)
  • Idempotency-Key support: fail, 0 of 3, about 1.5 points to gain. Ora saw: No REST Idempotency-Key support detected
  • Scoped permissions: partial, 2 of 5, about 1.5 points to gain. Ora saw: OpenAPI declares security schemes but no named OAuth scopes - agents get all-or-nothing access. Declare per-scope grants (e.g. read:*, write:*) in the spec.
  • Rate limit response headers: fail, 0 of 2, about 1 points to gain. Ora saw: No REST rate-limit headers found on probed endpoints
  • Content without JavaScript: partial, 2 of 3, about 0.5 points to gain. Ora saw: Homepage: 1009 chars with H1, but 1.0% content ratio is below the 5% target. 4 of 4 sampled inner pages have text without JavaScript (1 skipped).

Our read

Vercel scores 91 and a perfect Discovery layer, 21 of 21, and the act layer is nearly as complete: an MCP server at mcp.vercel.com that requires authentication, an OpenAPI 3.0.3 specification at vercel.com/openapi.json, OAuth authorization server metadata on the main domain, the vercel CLI on npm, an agent skills index with nine skills, an llms.txt and an agents.md served as text/markdown. What Ora marks down is finer than any of that: the specification declares security schemes but names no OAuth scopes, so an agent gets all-or-nothing access; the REST endpoints answer without rate-limit headers or an Idempotency-Key; and none of the 43 script bundles registers a WebMCP tool. The home page is 1,009 characters at a 1 percent text ratio, which is the one reading-side mark against it. This is the second-highest score in the directory's first sixty, and the fixes left are specification details.

Test your own surfaces

The scan above is free and needs no account: enter a domain at ora.ai and read the same four layers for your own site. For an MCP server, Ora scans the server URL itself and runs 28 of its 132 public checks on it, which is why it is the check this directory recommends; the reasons, the limits and the alternatives are on their own 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.

npx ax@0.10 audit https://example.com --min-score 70

All 360 checks, one company a day. Corrections to the corrections page.