Is Agentic scores a public website on what AI agents can discover, retrieve, understand and use, returning a number out of 100 from a scan run by Ora and reachable as a one-line command. Its published examples span a wide range: vercel.com at 86, mixpanel.com at 89, stripe.com at 74, github.com at 60, nytimes.com at 44 and wordpress.com at 38.
The scoring design is more interesting than the leaderboard. Essential checks carry most of the weight and cover fundamentals rather than novelties: server-rendered content, correct HTTP behaviour, clear document structure, recoverable errors and usable controls. Recommended checks activate only when scan evidence shows the site actually offers an API, OAuth flow, GraphQL endpoint, MCP server, developer portal or commerce surface, and checks that do not apply are excluded rather than counted as failures. That means a site is measured against what it claims to be, so a content site is not penalised for lacking an API.
Ranking the basics first matches what agents actually break on, which is the operational reading in agents searching, fetching and browsing as separate operations. The agents hub tracks the machine-readable side of the web.
The scores come from one vendor's scanner with its own rubric, and the featured numbers are the vendor's selection rather than a sampled population. The condition to watch is whether the rubric and per-check weights are published, because a readiness score without a visible rubric measures agreement with the scanner rather than readiness.