Websites now get graded for agent readiness. Vercel has released Is Agentic, a free public tool that scores how readily AI agents can discover, access, understand, and use a website. The release arrived on August 23, 2026, with scans run and scored by Ora, an agent-experience research company from era labs.
Vercel operates the interface, report pages, storage, and the grouping that produces the displayed score. Is Agentic has no paid plans, subscription charges, or per-report fees, and its public site, read-only API, CLI, and MCP server require no API key or billing account.
Anyone with a public web surface can use the tool. Enter a URL in the browser, or run npx is-agentic <domain> for a standard result or npx is-agentic <domain> –json when the output needs to feed another system. Seed-stage startups get a free baseline audit without procurement, mid-market SaaS teams can use the JSON output as a CI regression gate, and enterprises can benchmark documentation portals, developer sites, and commerce surfaces across business units.
Scoring Websites Like Agents Actually Use Them
Ora’s methodology follows four layers an agent moves through: Discovery, Access, Usability, and Payments. Discovery carries 20 points across 15 checks, Access carries 30 points across 41 checks, Usability carries 40 points across 56 checks, and Payments carries 10 points across 6 checks.
That adds up to 118 checks, matching Vercel’s “100+ checks” claim. Ora says its checklist was reverse-engineered from real agent runs rather than authored from opinion—a sensible distinction when a polished checklist can otherwise become a museum of assumptions.
Ora publishes a letter scale: A+ covers 95–100, A covers 86–94, B covers 70–85, C covers 48–69, D covers 28–47, and F covers 0–27. Vercel regroups the checks into its own displayed scoring model, with essential checks sharing an 80-point pool, recommended checks sharing 20 points, and emerging signals adding a bonus capped at five points.
Missing emerging signals never lowers a score, while not-applicable checks are excluded instead of counted as failures. Partial results receive proportional credit, and duplicated check IDs across MCP surfaces are averaged. Recommended checks activate only when scan evidence positively identifies an API, OAuth flow, GraphQL endpoint, MCP server, developer portal, or commerce surface, so a marketing site is not penalized for omitting an interface it never claimed to offer.
Every finding includes observed evidence and, where available, a concrete recommendation. Real check IDs include content-no-js, agent-friendly-404, markdown-negotiation-vary, json-ld, sitemap, trust-anchors, and metadata-completeness; the 404 check demands a real HTTP 404 status rather than a 200 response serving an app shell.
Reports Built for Humans, Agents, and CI
A live API response exposes the score, score_label, scanned_at, eligible_checks, and a score_breakdown with earned and available points for each tier. It also includes an issues array, with every issue carrying an id, name, details, recommendation, result, and tier.
Reports include an observed agent journey showing where one agent encountered friction, but Vercel keeps that run outside the numeric score. The separate Prompt to fix feature turns actionable recommendations into an implementation brief for a coding agent, because apparently even repair work now needs a handoff document.
Three read-only surfaces expose completed reports. The report API lives at /api/v1/report and allows 120 requests per client IP per 60-second window. Errors use RFC 9457 problem details with stable codes such as invalid_url, report_not_found, and rate_limit_exceeded.
The OpenAPI description serves as the supported integration contract and is discoverable through an RFC 9727 API catalog. Deprecations use RFC 9745 headers, with Sunset announced at least 90 days ahead. The MCP server at /mcp exposes is_agentic_get_report, is_agentic_get_methodology, and is_agentic_get_developer_docs over Streamable HTTP; hosts that support MCP Apps can render an interactive score card.
An official agent skill installs through npx skills add vercel-labs/is-agentic. Is Agentic also practices what it scores: report pages render the score in the initial HTML response, serve a Markdown variant under Accept: text/markdown, and set Vary: Accept so shared caches do not mix representations.
