ora, together with Vercel, Stripe and others, releases the AgentReady open standard
English text is being prepared.
ora, together with Vercel, Stripe, Auth0, Mintlify, Tavily, Google, Elementor and MIT, has launched the AgentReady specification, open-sourced under the MIT license, which defines the extent to which websites must be built so that AI agents can find, understand and operate them. Its research team ran agents on real tasks more than 250,000 times; controlled experiments using 19 configurations and 190 probes in total found that only hiding answers in JavaScript and returning 403 to agents caused failures, while llms.txt, JSON-LD, sitemaps and Markdown mirrors did not change the results. The specification sets out MUST, SHOULD and MAY requirements across the three steps of find, read and operate, with the operate layer requiring at least one of an MCP server or a CLI.
Why read this
AI score 73
An open specification based on 250,000 real-world tests that identifies the few conditions under which agents actually fail when accessing websites, serving as a reference for site improvements.
Topics
View original postIntroducing https://t.co/ofzXbW4lOc, the first open standard for agent experience. Agents are a new class of user, but there's still no shared definition of what it takes to be ready for them. So we partnered with the leading companies in the agentic web such as Vercel, Stripe, Auth0, Mintlify, Tavily and Elementor. Our research lab ran agents on real tasks more than 250,000 times to measure what actually decides whether an agent succeeds, and every requirement traces back to that data. The result is a short set of essentials in the order agents meet a product: find, read, act. We've been overwhelmed by requests to join. If you build or study the surfaces agents touch, come help shape it. The spec: https://t.co/MuSWchN40k The story behind it: https://t.co/h1kBIsjlXM
View original
Introducing https://t.co/ofzXbW4lOc, the first open standard for agent experience. Agents are a new class of user, but there's still no shared definition of what it takes to be ready for them. So we partnered with the leading companies in the agentic web such as Vercel, Stripe, Auth0, Mintlify, Tavily and Elementor. Our research lab ran agents on real tasks more than 250,000 times to measure what actually decides whether an agent succeeds, and every requirement traces back to that data. The result is a short set of essentials in the order agents meet a product: find, read, act. We've been overwhelmed by requests to join. If you build or study the surfaces agents touch, come help shape it. The spec: https://t.co/MuSWchN40k The story behind it: https://t.co/h1kBIsjlXM
Body
AgentReady 发布：基于 25 万次 Agent 实测的 "Agent 可用性" 开放标准，真正卡住 Agent 的只有两件事
@oradotai 联合 Vercel、Stripe、Auth0、Mintlify、Tavily、Google、Elementor、MIT 推出 AgentReady：一份开源（MIT 许可）、不绑定任何厂商的规范，规定网站和产品要做到什么程度，AI Agent 才能找到它、读懂它、用起来。核心发现很反直觉：在受控实验里，只有两种改动会让 Agent 失败：把答案藏在 JavaScript 里，以及直接屏蔽 Agent。其他十几种 "给 Agent 优化" 的配置，对能不能拿到答案都没有影响。 https://t.co/YEnIQ5ZJwA