Lyrenth is an independent web index and reading API built for AI. Language models don't need HTML. They need the content: the article, the docs page, the release notes, without navigation, ads, cookie banners, and markup noise. Lyrenth fetches any URL and returns a clean, structured document in Markdown, with the title, headings, and structure intact, ready to drop straight into a prompt or a pipeline. This matters for two reasons. Quality: models reason better over clean text than over raw HTML. Cost: pages shrink dramatically once the noise is gone, so every read spends fewer tokens. Behind the API is our own index of the public web, over 2.8 billion documents and growing. It is built by AIWebIndex, our verified crawler: published IP addresses, cryptographically signed requests, robots.txt honored. Lyrenth is built for developers shipping agents, assistants, RAG pipelines, and research tools. Use the REST API directly, or connect it over MCP so agents like Claude can read the web natively. Everything in it is live and verifiable today: the counter is public at lyrenth.com/web-index/stats, the crawler verification is documented by three independent bot directories, and the MCP integration is listed in the Claude connectors directory.