x402 Fast PDF Parser v0.1.0 (Base mainnet)

Pay-per-use PDF text extraction over the x402 payment protocol.
Born-digital PDFs only (no OCR). Speed is the product: every response
includes parse_ms so you can hold us to it.

ENDPOINT
  POST /parse
    Content-Type:      application/pdf   (raw PDF bytes, max 10MB)
    X-Document-SHA256: hex SHA-256 of the uploaded bytes (required)
  -> 200 JSON: { "text": "...", "pages": 12, "parse_ms": 38, "sha256": "..." }

PRICE
  $0.002 USDC per document on Base mainnet (eip155:8453). This is REAL
  money — every parse settles an on-chain USDC transfer.

HOW TO PAY (agents)
  1. POST your document. You'll get HTTP 402 with x402 payment requirements.
  2. Sign the payment with any x402 v2 client (e.g. @x402/fetch wrapping fetch,
     funded by a Base mainnet USDC wallet) and retry with the payment header.
  3. The response settles on-chain automatically; the receipt is in the
     PAYMENT-RESPONSE header.

NOTE on parse_ms: Cloudflare freezes in-request clocks during CPU work
(Spectre mitigation), so parse_ms reads ~0 in production at the edge. It is
accurate in local dev and will be accurate on the heavy (Lambda) path.

RULES
  - One payment proof per document: replays within 10 minutes get 409.
  - 50 requests/hour per paying wallet, then 429.
  - Unpaid request floods are throttled per-IP at the edge (100/min), then 429.
  - Malformed/unparseable PDFs get 422 (you are not charged: payment only
    settles after a successful parse).

PRIVACY
  No document retention: PDF bytes are parsed in memory and never written to storage or logs. We retain only a SHA-256 fingerprint of the document (replay protection, 10-min TTL) and the payer address (rate limiting, 1-hour TTL). Document content is not logged by the application.

OTHER ROUTES
  GET /health                  — free liveness check.
  GET /.well-known/x402.json   — free machine-readable service manifest.
  GET /llms.txt                — free plain-text description for AI crawlers.