{
  "format": "scrubnet-page",
  "version": 1,
  "language": "en-GB",
  "title": "Markdown for Agents and the Machine-Readable Web | Scrubnet",
  "description": "What Cloudflare’s Markdown for Agents proposal reveals about low-noise formats, crawler access and machine-readable publishing.",
  "source": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html",
  "representations": {
    "html": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html",
    "markdown": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.md",
    "text": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.txt",
    "json": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.json"
  },
  "languageAlternates": {
    "en-GB": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html",
    "fr-FR": "https://scrubnet.org/fr/articles/markdown-agents-web-lisible-machines.html"
  },
  "content": [
    {
      "type": "section",
      "label": "Article hero",
      "content": [
        {
          "type": "heading",
          "level": 1,
          "content": [
            {
              "type": "text",
              "text": "Markdown for Agents"
            }
          ]
        },
        {
          "type": "paragraph",
          "content": [
            {
              "type": "text",
              "text": "What Cloudflare’s proposal means for machine-readable publishing"
            }
          ]
        }
      ]
    },
    {
      "type": "aside",
      "label": "Article author",
      "content": [
        {
          "type": "paragraph",
          "content": [
            {
              "type": "image",
              "url": "https://scrubnet.org/charles-vayssie.webp",
              "alt": "Charles Vayssié, founder of Scrubnet"
            }
          ]
        },
        {
          "type": "paragraph",
          "content": [
            {
              "type": "text",
              "text": "By "
            },
            {
              "type": "link",
              "url": "https://scrubnet.org/about-us.html",
              "content": [
                {
                  "type": "text",
                  "text": "Charles Vayssié"
                }
              ]
            }
          ]
        },
        {
          "type": "paragraph",
          "content": [
            {
              "type": "text",
              "text": "Founder, Technical SEO & Web Developer"
            }
          ]
        },
        {
          "type": "paragraph",
          "content": [
            {
              "type": "text",
              "text": "Technical SEO specialist and web developer leading strategy and delivery across complex e-commerce and international websites, with expertise in migrations, rendering, crawl management, structured data and Core Web Vitals."
            }
          ]
        }
      ]
    },
    {
      "type": "article",
      "content": [
        {
          "type": "section",
          "id": "intro",
          "content": [
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "link",
                  "url": "https://blog.cloudflare.com/markdown-for-agents/",
                  "content": [
                    {
                      "type": "text",
                      "text": "Cloudflare recently proposed “Markdown for Agents”"
                    }
                  ]
                },
                {
                  "type": "text",
                  "text": " as a cleaner publishing format designed specifically for AI systems."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "The core idea is simple. Large language models do not need decorative layout, client side rendering logic, cookie banners, or visual hierarchy tricks. They need structured, predictable, low noise content."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Markdown is compact, readable, and deterministic. It removes presentation clutter and exposes the informational layer of the page."
                }
              ]
            }
          ]
        },
        {
          "type": "figure",
          "image": {
            "url": "https://scrubnet.org/discoveries/cloudflare-markdown-scrubnet.webp",
            "alt": "Scrubber duck simnplifying content for crawlers"
          },
          "caption": [
            {
              "type": "text",
              "text": "Scrubberduck in the digital world simplifying content for crawlers."
            }
          ]
        },
        {
          "type": "section",
          "id": "problem",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "The underlying problem"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "The modern web is optimised for humans using browsers. It assumes:"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "JavaScript execution"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "CSS driven layout"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Dynamic hydration"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Advertising and tracking layers"
                        }
                      ]
                    }
                  ]
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "For agents, this is overhead. Every extra byte increases crawl cost. Every rendering dependency introduces uncertainty. Every layout artifact becomes noise."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "AI systems do not “see” pages. They ingest text streams and structured hints. The cleaner the stream, the better the understanding."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "cloudflare",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "What Markdown for Agents represents"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Cloudflare’s proposal signals a shift in thinking. Instead of forcing agents to interpret the human web, publishers can expose a machine friendly representation."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Markdown provides:"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Clear heading hierarchy"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Minimal markup"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Deterministic content boundaries"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Reduced token waste for LLM ingestion"
                        }
                      ]
                    }
                  ]
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "It separates information from decoration. That separation is foundational."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "scrubnet-alignment",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "Why this aligns with the Scrubnet Observatory"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Scrubnet’s crawler observatory was built on the same premise. Machines need their own layer of the web."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Its feeds already provide:"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Clean HTML without rendering dependency"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "JSON representations for structured ingestion"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Plain text versions for deterministic parsing"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Timestamped, machine first publishing"
                        }
                      ]
                    }
                  ]
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Markdown for Agents is another expression of the same idea. Reduce noise. Remove guesswork. Publish content in a format that agents can consume fully in one pass."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "economics",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "The economics of machine consumption"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "LLMs operate under token constraints. Crawlers operate under bandwidth and compute constraints. Rendering is expensive. Parsing noisy DOM trees is inefficient."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "A compact, linear representation of content reduces:"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Token waste during ingestion"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Ambiguity in structure"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Dependency on JavaScript execution"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Inference errors caused by layout artifacts"
                        }
                      ]
                    }
                  ]
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Clean publishing is not aesthetic minimalism. It is operational efficiency."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "layers",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "The web is splitting into layers"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "One layer serves humans. It is interactive, visual, expressive."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Another layer serves machines. It is structured, bounded, and explicit."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Cloudflare’s proposal is further evidence that machine-readable representations are becoming an explicit part of web publishing. Scrubnet’s observatory examines how crawlers interact with several such representations."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "difference",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "Markdown is a format. The observatory is infrastructure."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Markdown for Agents focuses on representation. The Scrubnet observatory provides public representations and observes their discovery and request patterns."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "The observatory adds:"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Brand level feed architecture"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Dedicated agent endpoints"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Bot consumption tracking"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Deterministic update signals"
                        }
                      ]
                    }
                  ]
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Format alone does not create a machine web. Structured publishing plus measurable ingestion does."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "future",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "What this signals"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Major infrastructure players are now acknowledging that AI agents require explicit support."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "This is not a trend. It is a structural shift."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "The web is no longer only read by humans and indexed by search engines. It is actively consumed by generative systems that reuse, summarise, and reason over content."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Machine first publishing is becoming a baseline requirement."
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "takeaways",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "Key takeaways"
                }
              ]
            },
            {
              "type": "list",
              "ordered": false,
              "items": [
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "AI agents benefit from low noise, structured formats"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Markdown reduces rendering and parsing overhead"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Machine specific representations are becoming standard"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Scrubnet operationalises the machine web layer"
                        }
                      ]
                    }
                  ]
                },
                {
                  "content": [
                    {
                      "type": "paragraph",
                      "content": [
                        {
                          "type": "text",
                          "text": "Clean, bounded feeds are aligned with agent economics"
                        }
                      ]
                    }
                  ]
                }
              ]
            }
          ]
        },
        {
          "type": "section",
          "id": "cta",
          "content": [
            {
              "type": "heading",
              "level": 2,
              "content": [
                {
                  "type": "text",
                  "text": "Make machine-readable access observable"
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "text",
                  "text": "Scrubnet helps authorised participating sites expose structured machine-readable representations and observe verified crawler requests. Access does not by itself prove indexing, ingestion or downstream use."
                }
              ]
            },
            {
              "type": "paragraph",
              "content": [
                {
                  "type": "link",
                  "url": "https://scrubnet.org/brands.html",
                  "content": [
                    {
                      "type": "text",
                      "text": "Add a Site"
                    }
                  ]
                },
                {
                  "type": "text",
                  "text": " "
                },
                {
                  "type": "link",
                  "url": "https://scrubnet.org/llms.html",
                  "content": [
                    {
                      "type": "text",
                      "text": "Explore Public Feeds"
                    }
                  ]
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "structuredData": [
    {
      "@context": "https://schema.org",
      "@graph": [
        {
          "@type": "BlogPosting",
          "@id": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html#article",
          "mainEntityOfPage": {
            "@type": "WebPage",
            "@id": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html"
          },
          "headline": "Markdown for Agents and the Machine-Readable Web",
          "alternativeHeadline": "Markdown Is Not About Style. It Is About Machines",
          "description": "What Cloudflare’s Markdown for Agents proposal reveals about low-noise formats, crawler access and machine-readable publishing.",
          "image": "https://scrubnet.org/discoveries/cloudflare-markdown-scrubnet.webp",
          "datePublished": "2026-02-22",
          "dateModified": "2026-07-22",
          "author": {
            "@id": "https://scrubnet.org/#founder"
          },
          "publisher": {
            "@id": "https://scrubnet.org/#org"
          },
          "inLanguage": "en-GB"
        },
        {
          "@type": "BreadcrumbList",
          "@id": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html#breadcrumbs",
          "itemListElement": [
            {
              "@type": "ListItem",
              "position": 1,
              "name": "Home",
              "item": "https://scrubnet.org/"
            },
            {
              "@type": "ListItem",
              "position": 2,
              "name": "Articles",
              "item": "https://scrubnet.org/discoveries.html"
            },
            {
              "@type": "ListItem",
              "position": 3,
              "name": "Markdown for Agents and the Machine-Readable Web",
              "item": "https://scrubnet.org/discoveries/markdown-for-agents-machine-web.html"
            }
          ]
        },
        {
          "@type": "Organization",
          "@id": "https://scrubnet.org/#org",
          "name": "Scrubnet Ltd",
          "url": "https://scrubnet.org/",
          "logo": {
            "@type": "ImageObject",
            "url": "https://scrubnet.org/scrubberduck-72.webp"
          }
        },
        {
          "@type": "Person",
          "@id": "https://scrubnet.org/#founder",
          "name": "Charles Vayssié",
          "jobTitle": "Founder, Technical SEO Specialist and Web Developer",
          "image": "https://scrubnet.org/charles-vayssie.webp",
          "url": "https://scrubnet.org/about-us.html",
          "worksFor": {
            "@id": "https://scrubnet.org/#org"
          },
          "sameAs": [
            "https://www.linkedin.com/in/charles-vayssie/",
            "https://x.com/Scrubnet_"
          ]
        },
        {
          "@type": "WebSite",
          "@id": "https://scrubnet.org/#website",
          "url": "https://scrubnet.org/",
          "name": "Scrubnet",
          "publisher": {
            "@id": "https://scrubnet.org/#org"
          },
          "inLanguage": "en-GB"
        }
      ]
    }
  ]
}