Vercel launches free Is Agentic audit tool for AI agents

Vercel and the agent-experience research company Ora have released Is Agentic, a free public tool that evaluates how readily artificial intelligence agents can discover, access, understand, and use a given website. The service runs automated scans that assign a numeric score based on more than a hundred checks covering four functional layers. Anyone can enter a URL in a browser or invoke the command-line interface to receive a detailed report without paying a fee or providing billing information. The launch signals a move toward standardized measurement of agent-friendly web properties.

Ora's methodology divides the evaluation into discovery, access, usability and payments. Discovery carries twenty points across fifteen checks, access contributes thirty points across forty-one checks, usability adds forty points across fifty-six checks, and payments contributes ten points across six checks, yielding a maximum possible total of 118 checks. Each layer reflects a stage an agent typically traverses when interacting with a web surface, from first locating a page to completing a transaction. The check count matches Vercel's claim of "100+ checks", confirming the breadth of the audit.

Vercel regroup Ora's checks into its own displayed scoring model. Essential checks occupy an eighty-point pool, recommended checks a twenty-point pool, and emerging signals may add a bonus capped at five points; their absence never lowers a score. Checks that do not apply to a scanned site are excluded rather than counted as failures. When a scan returns partial results, each check earns proportional credit, and duplicate check identifiers across multiple MCP surfaces are averaged. This logic ensures that a marketing-only site is not penalized for omitting interfaces it never claimed to offer.

Every finding ships with observed evidence and, where available, a concrete recommendation. A live API response returns the score, a label such as A-plus through F, the timestamp of the scan, the list of eligible checks, a breakdown of earned and available points per tier, and an issues array. Each issue records an identifier, a human-readable name, details of the observed problem, a recommendation for remediation, the result of the check, and its tier. Real check IDs include content-no-js, agent-friendly-404, markdown-negotiation-vary, json-ld, sitemap, trust-anchors, and metadata-completeness. The 404 check, for example, requires a genuine HTTP 404 status rather than a 200 response serving an app shell.

Completed reports are exposed through three read-only surfaces. The report API lives at /api/v1/report and is rate-limited to 120 requests per client IP within a sixty-second window. Errors follow RFC 9457 problem-details format with stable codes such as invalid_url, report_not_found, and rate_limit_exceeded. The OpenAPI description serves as the supported integration contract, discoverable through an RFC 9727 API catalog. Deprecations are signalled via RFC 9745 headers, with a minimum ninety-day sunset notice. An MCP server at https://is-agentic.com/mcp publishes three methods is_agentic_get_report, is_agentic_get_methodology and is_agentic_get_developer_docs over streamable HTTP, allowing hosts that support MCP Apps to render an interactive score card. An official agent skill can be installed via the command npx skills add vercel-labs/is-agentic.

Is Agentic is designed for a wide range of organisations. Seed-stage startups receive a free baseline audit without procurement overhead, while mid-market SaaS teams can pipe the -json output into continuous-integration pipelines as a regression gate. Enterprises can benchmark documentation portals, developer sites, and commerce surfaces across business units. Target industries include developer tools and SaaS, e-commerce and retail, travel and hospitality, fintech, marketplaces, healthcare provider directories, and media or documentation publishers. Typical applications encompass pre-launch agent-readiness audits, CI checks that fail builds when server-rendered content regresses, competitive benchmarking against public scores, validation of an MCP server or OpenAPI surface, and prioritisation of technical fixes with copy-pasteable remediation prompts.

The launch of Is Agentic introduces a free, standardized way for teams to gauge how friendly their web properties are to autonomous agents, a growing concern as AI-driven assistants become more prevalent in commerce, support and workflow automation. By exposing concrete check IDs and remediation prompts, the tool lowers the barrier for developers to improve discoverability, accessibility and transaction capability. Potential limitations include reliance on a single agent-run simulation, the possibility that emerging agent behaviours are not yet captured, and the need for organisations to interpret scores in the context of their specific use cases. Watchers should monitor future updates to the check list, changes to the scoring thresholds, and any expansion of paid or enterprise-grade features that Vercel may introduce.