Cloudflare
@Cloudflare
We have recently added "Docs for agents" in our docs homepage to help you use your favourite AI agents to consume Cloudflare documentation. https://t.co/9cJ5ehjX6K https://t.co/ir4s13q0CD
19retweets198likes
View on XCloudflare Launches Docs for Agents to Standardize Machine Readable Documentation
Cloudflare, a network and security platform, launched Docs for agents, a documentation framework providing machine-readable content for AI tools like Cursor and GitHub Copilot. The system includes site-wide
llms.txt files and supports content negotiation (a mechanism where a client requests a specific format) to serve Markdown instead of HTML.This update implements the Cloudflare Agent Readiness Score standards to solve the "token tax" of messy web scraping. By providing clean Markdown and managed MCP (Model Context Protocol — an open standard for tool access) servers, Cloudflare reduces context window overhead while improving response accuracy.
You can now connect AI agents directly to Cloudflare’s managed remote MCP servers to search documentation or call over 2,500 API endpoints. Developers can also install task-specific Agent Skills to teach agents how to deploy Workers. Every documentation page is now accessible as Markdown by appending /index.md.
Still wondering? A few quick answers below.
It is a documentation hub designed for AI tools like Cursor and GitHub Copilot. Instead of standard HTML, it provides content in agent-friendly formats like Markdown and through the Model Context Protocol. This allows agents to answer questions, generate configurations, and call Cloudflare APIs with higher accuracy and lower token usage.
You can access Markdown versions of any page by appending /index.md to the URL or by sending an Accept: text/markdown HTTP header. Cloudflare also provides site-wide and per-product llms.txt files, which follow an open standard for providing structured, machine-readable indexes that AI models can easily crawl and process.
Agent Skills are structured, task-specific instructions that teach AI agents how to perform complex Cloudflare operations. These skills cover products like Workers, security, and networking. By loading these instructions from the official Cloudflare Skills repository, an agent gains specialized knowledge on how to deploy resources or configure firewall rules that it would not otherwise have.
Cloudflare provides managed remote servers using the Model Context Protocol, an open standard for connecting AI tools to external data. You can use Code Mode for broad access to over 2,500 API endpoints or domain-specific servers for documentation and analytics. These servers allow your agent to search docs and interact with services directly.
Yes, the documentation hub and its associated machine-readable endpoints are publicly available. Developers can access the Markdown content, OpenAPI specifications, and the open-source repositories for Agent Skills and MCP servers. Specific setup guides are provided for popular AI agents to help users integrate these resources into their existing development workflows.
