Vercel Introduces 'Is Agentic', a Free Agent-Readiness Scoring Tool That Audits Public Websites Using Ora's 100+ Checks
Vercel launched "Is Agentic," a free public tool that scores website agent-readiness across four layers: discovery, access, usability, and payments The scoring methodology, developed by Ora (era labs), comprises 118 checks reverse-engineered from real agent runs rather than opinion-based authoring Vercel regroups Ora's checks into a weighted model: 80 points for essential checks, 20 for recommended, and up to 5 bonus points for emerging signals The tool is accessible via browser, CLI (`npx is-ag
Analysis
TL;DR
- Vercel launched "Is Agentic," a free public tool that scores website agent-readiness across four layers: discovery, access, usability, and payments
- The scoring methodology, developed by Ora (era labs), comprises 118 checks reverse-engineered from real agent runs rather than opinion-based authoring
- Vercel regroups Ora's checks into a weighted model: 80 points for essential checks, 20 for recommended, and up to 5 bonus points for emerging signals
- The tool is accessible via browser, CLI (
npx is-agentic <domain>), read-only API, and MCP server — all free with no API key or billing required - Reports include concrete evidence, actionable recommendations, and a "Prompt to fix" feature that generates implementation briefs for coding agents
Why It Matters
As AI agents increasingly act on behalf of users — shopping, booking, integrating, and citing information — the discoverability and accessibility of public web surfaces becomes a critical infrastructure concern. This tool provides the industry's first standardized, free scoring framework for agent-readiness, enabling organizations to proactively audit and improve their web properties for agent consumption rather than human-only interaction.
Technical Details
- Scoring Methodology: Ora's framework walks four agent interaction layers — Discovery (20 pts, 15 checks), Access (30 pts, 41 checks), Usability (40 pts, 56 checks), and Payments (10 pts, 6 checks) — totaling 118 checks reverse-engineered from real agent runs
- Letter Grade Scale: A+ (95-100), A (86-94), B (70-85), C (48-69), D (28-47), F (0-27), with Vercel regrouping into Essential (80-point pool), Recommended (20-point pool), and Emerging signals (5-point bonus cap)
- Applicability Logic: Recommended checks activate only when scan evidence positively identifies an API, OAuth flow, GraphQL endpoint, MCP server, developer portal, or commerce surface; marketing sites are not penalized for missing interfaces they never claimed to offer
- Machine Interfaces: Read-only report API at
/api/v1/report(120 requests per IP per 60s), OpenAPI description via RFC 9727 catalog, MCP server exposingis_agentic_get_report,is_agentic_get_methodology, andis_agentic_get_developer_docsover Streamable HTTP - Real Check Examples:
content-no-js,agent-friendly-404(requires real HTTP 404 rather than 200 app shell),markdown-negotiation-vary,json-ld,sitemap,trust-anchors,metadata-completeness; reports practice what they score with server-rendered HTML, Markdown variant underAccept: text/markdown, and properVary: Acceptheaders
Industry Insight
- Organizations should integrate agent-readiness audits into pre-launch workflows and CI pipelines (using
--jsonoutput as a regression gate) to prevent agent-accessibility degradation as sites evolve - The tool is particularly relevant for developer tools/SaaS, e-commerce, travel, fintech, and healthcare — any sector where AI agents may act on behalf of users — and mid-market teams can wire automated checks directly into their deployment pipelines
- Vercel explicitly positions this as a prioritized technical review, not a certification, security audit, or accessibility review; professionals should treat scores as directional guidance for technical remediation rather than compliance validation
Disclaimer: The above content is generated by AI and is for reference only.