Cloudflare
@Cloudflare
Today, we are excited to introduce https://t.co/F5gmrAYGFP — a new tool to help site owners understand how they can make their sites optimized for agents. https://t.co/2xAeZlX5AI
123retweets916likes
View on XCloudflare Launches Agent Readiness Score to Standardize the Machine Readable Web
· Updated
Cloudflare launched the Agent Readiness score and a public testing tool to help site owners optimize for autonomous AI. The scanner audits URLs for discoverability, content accessibility, bot access control, and capabilities like
MCP (an open standard for connecting models to tools). It provides prompts to help implement missing standards.Most websites are built for human eyes, forcing agents into a grep loop where they waste tokens searching messy HTML. This follows Cloudflare's recent move to serve structured error responses to agents. Shifting to machine-friendly formats like Markdown helps bypass context window limits that often cause agentic workflows to fail.
You can scan any domain to receive a score and technical roadmap for improvement. The tool checks for emerging protocols like x402 and Agent Skills discovery. This builds on redirecting training crawlers to ensure agents ingest canonical content, with checks integrated into the Cloudflare URL Scanner API for automated auditing.
Still wondering? A few quick answers below.
The Agent Readiness score is a diagnostic metric from Cloudflare that measures how well a website supports AI agents. It evaluates sites across four categories: discoverability, content accessibility, bot access control, and protocol capabilities. The goal is to help site owners transition from a human-centric web to a machine-readable web that autonomous AI can navigate efficiently.
When you enter a URL into isitagentready.com, Cloudflare makes several requests to the site to check for specific technical standards. It looks for files like robots.txt and sitemaps, tests if the server can provide Markdown content instead of HTML, and checks for protocol discovery endpoints like Model Context Protocol server cards or API catalogs.
The scanner checks for several emerging and established standards, including robots.txt for crawl rules, Markdown content negotiation for token efficiency, and the Model Context Protocol for tool use. It also audits for Content Signals to define AI training preferences, Web Bot Auth for secure identification, and commerce protocols like x402 for machine-to-machine payments.
Developers can use the public web interface at isitagentready.com or access the functionality through the Cloudflare URL Scanner. For automated workflows, the checks are available via the URL Scanner API by passing a specific agentReadiness option in the scan request. This allows teams to programmatically audit their sites for AI compatibility at scale using their existing API tokens.
Optimizing for agents reduces the grep loop, where AI wastes resources repeatedly searching for keywords in messy HTML. Cloudflare found that agent-friendly sites allow AI tools to arrive at correct answers 66% faster while consuming 31% fewer tokens. This makes interactions significantly cheaper and more reliable by ensuring content fits within the agent's limited context window.
