Read the web as Markdown.

Any URL in, clean Markdown out, with an AIO score telling you how readable that page is to an agent. We honour robots.txt, and we tell you when a read came back thin instead of handing you an empty page.

Quick start

$ curl https://api.buildonto.dev/v1/read \
    -H "Authorization: Bearer onto_sk_live_..." \
    -H "Content-Type: application/json" \
    -d '{"url": "https://stripe.com/pricing"}'

No key required

The reader is open. Put any URL after the host and read it straight away — md.buildonto.dev/example.com. Plain text for agents, a readable page in a browser. Same URL either way.

Endpoints

Method	Path	Returns
POST	/v1/read	Any URL to clean Markdown
POST	/v1/score	AIO score — how readable a page is to an agent
POST	/v1/read-and-score	Both, in one round trip
POST	/v1/batch	Many URLs in one call
POST	/v1/map	Discover the URLs on a site
POST	/v1/extract	Structured data from a page
GET	/v1/usage	Your quota and consumption
GET	/mcp	Hosted MCP endpoint for Claude and ChatGPT

Errors

Code	HTTP	Meaning
UNAUTHORIZED	401	Missing or invalid API key
INVALID_URL	400	Unparseable, non-http, or a private address
URL_NOT_FOUND	404	Target 404ed, or the host does not resolve
ROBOTS_BLOCKED	403	The site disallows AI crawlers — we honour that
WAF_BLOCKED	403	The site refused our crawler
UNSUPPORTED_TYPE	415	Not a document — image, video, archive
TOO_LARGE	413	Body exceeds the read limit
IMAGE_PDF	422	Scanned PDF with no extractable text
TLS_ERROR	502	The site's certificate is invalid
TIMEOUT	504	The site took too long
RATE_LIMITED	429	Monthly quota exceeded
PAYMENT_REQUIRED	402	Quota exceeded and credits are empty