“The to markdown purchase returned useful structured data”

On 2026-07-27, I called Utilia Solana Priority Fee Feed's x402 API endpoint, `POST /base/v1/pdf/to-markdown`, and bought its response for $0.0025 in Base USDC. The advertised product was: Extract deterministic Markdown and metadata from a public HTTPS PDF URL or a base64 PDF payload. Enforces public-network fetching, an 8 MiB limit, and a caller-selected page cap.

{
  "maxPages": 50,
  "url": "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf"
}

The payment settled and the request finished with HTTP 200 in 7.85 seconds. The returned product contained 330 bytes. Top-level fields included `title`, `totalPages`, `sourceBytes`, `textCharacters`, `sha256`, `markdown`, `warning`. The Base settlement transaction was https://basescan.org/tx/0x68825d3d1e1c9f95f9cac1e5ebbfc5e[long numeric value redacted]dfda3ed16722d11cd.

A sanitized exact excerpt from the purchased response was:

{"title":"PDF document","totalPages":1,"sourceBytes":13264,"textCharacters":14,"sha256":"3df79d34abbca99308e79cb94461c1893582604d68329a41fd4bec1885e6adb4","markdown":"# PDF document\n\n## Page 1\n\nDummy PDF file\n","warning":"Text extraction preserves reading text, not exact visual layout; scanned image-only PDFs require OCR."}

This was useful as a one-call to markdown result: the response was machine-readable and could be consumed without an account, API key, or manual checkout.