Support
The docs describe the endpoints, schemas, authentication policy and OpenAPI description, and a methodology page explains the scoring.
Learn moreIs Agentic by Vercel
Scores how ready a website is for AI agents, with evidence
Intermediate API
Own Is Agentic by Vercel? Verify ownership to get a dofollow link
What is Is Agentic by Vercel?
Is Agentic by Vercel is an analytics tool. Is Agentic scans a public URL and returns a 0 to 100 score for how easily AI agents can discover, access and use the site, with per-check evidence, recommendations, and report access through a JSON API and MCP server.
Is Agentic is a web-based scanner that rates how well a public website can be discovered, read and used by AI agents. A visitor enters a URL and receives a score out of 100, along with a report that lists what was observed during the scan and, where possible, a concrete recommendation for fixing each gap. Every scan is run by Ora, and the service is offered under the Vercel name. The score is built from two groups of checks. Essential checks carry most of the weight and cover fundamentals: server-rendered content, correct HTTP behavior, clear document structure, recoverable errors, usable controls, and identity metadata such as a canonical URL, a declared language and Open Graph tags. Recommended checks only switch on when the scan finds signs of an API, OAuth flow, GraphQL endpoint, MCP server, developer portal or commerce surface, so a plain site is not penalized for lacking an interface it never offered. Emerging formats can add limited bonus credit, but missing them never lowers the result. In practice, the report is meant to be worked through from the top: failed Essential checks first, then Recommended gaps that match what the product actually does. Each report also includes an observed agent journey showing where one agent hit friction, kept outside the numeric score because a single task cannot represent every agent. After a fix ships, rescanning the same URL replaces the report with a fresh snapshot at a stable, shareable address. The tool is aimed at web developers, product teams and technical marketers who want a prioritized technical review of agent readiness. Reports can be read as HTML, as Markdown when an agent requests it, as JSON through a public API, or through an MCP server, none of which starts a scan or needs a credential. The homepage is explicit that this is not a guarantee of agent success, and that security, accessibility, privacy and legal compliance need their own reviews. Compared with general SEO or performance auditors, its focus is narrower and specific to agent access.
How do you use Is Agentic by Vercel?
1Enter your site URL
Open the homepage and type the public URL you want to evaluate into the Website URL field, then run the scan to get a score.
2Read the score report
Review the overall score and each check's evidence. Start with any failed Essential checks, since they describe the baseline an ordinary agent needs.3Compare against example scans
Open a featured or recent report, such as vercel.com, to see how scores, findings and recommendations are laid out for a well-scoring site.
4Check the scoring methodology
Read how Essential, Recommended and bonus checks are weighted so you can judge which gaps match your product's real capabilities.
5Apply fixes and rescan
Implement the recommendations, then rescan the same URL to replace the report with a fresh snapshot at the same shareable address.6Pull reports programmatically
Use the docs to find the public JSON API, OpenAPI description and MCP server for fetching completed reports in your own workflow.
Pros and cons
Pros
- Findings include the evidence seen during the scan plus copyable recommendations, which makes fixes easy to hand to a developerAI
- Recommended checks apply only when an API, MCP server or commerce surface is detected, so simple sites are not unfairly marked downAI
- Stable report URLs render the score in initial HTML and offer Markdown and JSON versions for agents and softwareAI
- Public JSON API and read-only MCP server expose completed reports without credentialsAI
- Published methodology and sample scores for well-known sites give useful benchmarks for comparisonAI
Cons
- Scans only cover public pages, so authenticated workflows and private areas of an app are not evaluatedAI
- A report is a snapshot of one moment, and status codes or agent protocols can change after the scanAI
- The observed agent journey uses a single agent and task, so it is illustrative rather than representativeAI
- Machine interfaces return completed reports only and cannot start a new scanAI
- Does not review security, accessibility, privacy or legal compliance, despite any compliance related framingAI
How much does Is Agentic by Vercel cost?
Integrations
Completed reports can be retrieved through a public JSON API with an OpenAPI description, and through the Is Agentic MCP server, which offers an optional interactive score card for hosts supporting MCP Apps.
Learn moreFeatures
Scores a URL out of 100 using Essential, Recommended and bonus checks, with evidence and recommendations per finding, an observed agent journey, rescans, stable report URLs, Markdown and JSON outputs, a public API and an MCP server.
Learn moreFrequently asked questions about Is Agentic by Vercel
Does Is Agentic by Vercel have an API?
Yes, Is Agentic by Vercel offers an API.How do you use Is Agentic by Vercel?
The walkthrough on this page covers 6 steps: 1. Enter your site URL 2. Read the score report 3. Compare against example scans 4. Check the scoring methodology 5. Apply fixes and rescan 6. Pull reports programmatically.What platforms does Is Agentic by Vercel support?
Is Agentic by Vercel is available on Web App.What does Is Agentic by Vercel integrate with?
Completed reports can be retrieved through a public JSON API with an OpenAPI description, and through the Is Agentic MCP server, which offers an optional interactive score card for hosts supporting MCP Apps.What are the limitations of Is Agentic by Vercel?
Scans only cover public pages, so authenticated workflows and private areas of an app are not evaluated. A report is a snapshot of one moment, and status codes or agent protocols can change after the scan. The observed agent journey uses a single agent and task, so it is illustrative rather than representative.
