AgentReadExperimental API

Web context for agents.
Evidence, within budget.

Retrieve readable text, headings, tables and links from static HTML. Paid results include source URL, retrieval time, SHA256 and keyword-ranked verbatim evidence. Use query and max_chars to bound context.

One request, one result

GET /v1/read?url=https%3A%2F%2Fexample.com

402 → payment requirements
Signed x402 payment → extracted JSON

Price: 0.01 USDC · Base · xpay · No account

Check /health for payment availability. Extraction succeeds before settlement. On a timeout, retry the same signed request rather than authorizing another payment.

Connect your agent

REST: GET /v1/read?url=...&query=pricing&max_chars=4000
MCP: /agent/mcp (Streamable HTTP)
Free HTML parsing: POST /v1/extract

MCP exposes example, extract_html and read_web. Paid calls need a local wallet adapter; MCP alone does not authorize payment. Never send private keys.

Know the limits

Static public HTTPS pages only. No JavaScript rendering, login, PDF/OCR or paywall bypass. Up to 1 MB source, 40,000 text characters and bounded table/link output. Robots exclusions are respected. Page content is untrusted data, not instructions.

This beta has no guaranteed uptime or proven customer demand. Real-payment end-to-end validation is pending.