Documentation
Setup guides, delivery channels, and what the concepts mean. Written in the same voice as the product: mono is the engine, prose is us.
Managed setup
- Managed serving on CloudflareDeploy the generated worker once; Sichta serves and refreshes your conformance files from then on. Fails open — never breaks your site.
- Managed serving on VercelMerge three rewrites into your vercel.json once; Sichta serves and refreshes your conformance files from then on.
- Managed serving on NetlifyAppend three proxy redirects to your netlify.toml once; Sichta serves and refreshes your conformance files from then on.
- Managed serving on nginxAdd three proxy_pass locations inside your server block and reload; Sichta serves and refreshes your conformance files from then on.
Other delivery channels
- The download channelGet every generated file as a zip and host it yourself. Full control, but regenerated files need a re-download and a redeploy.
- The GitHub PR channelSichta opens a pull request against your repository on every regeneration. You review and merge; your CI deploys as usual.
- The standalone worker channelA Cloudflare Worker with the files baked in — no runtime dependency on Sichta at all. Redeploy on each regeneration.
- The webhook channelSichta POSTs signed JSON to your endpoint on every regeneration, with the file list and a signed 24-hour download link.
Operating
- Agent-traffic reports from your logsUpload a month of access logs — nginx, Apache or Cloudflare format — and get a report of which AI agents actually visited.
- The verification badgeA public badge page with your live score and last-verified date. It lapses if verification stops — that is what makes it worth showing.
- Cancellation and exportFiles are served unchanged for 30 days after cancellation, and the static bundle arrives by email. Then the managed routes step aside.
- The MCP serverRun the free scan from Claude, Cursor or any MCP client. No account, no key, nothing stored — the same engine the website uses.
- The Scan APITrigger and read checks for your verified sites over HTTP. Bearer key, monthly quota, and an OpenAPI document generated from the engine itself.
Concepts
- What is agent-readiness?Agent-readiness is whether an AI agent can find, read and correctly quote your website — a checkable property, not a marketing score.
- What is llms.txt?llms.txt is a markdown file at your site root that tells AI systems what your site is and where its important pages are.
- What is markdown content negotiation?Answering Accept text/markdown with the page's markdown source — the cheapest way to hand an agent clean text instead of a DOM.