# PayPerScreen

> Pay-per-call sanctions, crypto-wallet, email, and IP risk
> screening API, priced in USDC and paid over [x402](https://x402.org): no
> account, no API key, no subscription. Every request is authenticated by a
> signed on-chain payment at the moment it's made, in USDC on Base, Polygon
> or Arbitrum. Built
> around the assumption that most real callers are agents, not humans
> clicking a button.

PayPerScreen checks names against the OFAC SDN, UN Security Council, EU and UK
sanctions lists (fuzzy match; the lists actually loaded are named in every
response) and crypto wallet addresses against OFAC's
sanctioned digital-currency address list (exact match). It also runs
deterministic trust/risk composites (no LLM calls anywhere in the scoring
path) against email addresses and IP addresses, each returning
a riskLevel (low/medium/high) and a redFlag when a severe signal fires. It
returns results and signals. It does not make compliance decisions for
you.

## Docs

- [Methodology](https://payperscreen.com/methodology): a complete,
  signal-by-signal list of what every check actually does, and what it
  doesn't check yet. Read this to judge whether a check is sufficient
  for your use case.
- [API reference](https://payperscreen.com/docs): endpoints, pricing, the
  x402 payment handshake, request/response shapes, and honest current
  coverage.
- [Lists we check](https://payperscreen.com/lists): live record counts and
  last-refreshed dates for every list currently loaded, not a marketing
  description, the actual database state.
- [Reference client](https://payperscreen.com/examples/agent-client.mjs): a
  standalone Node.js script an agent can run directly. It reads a private key
  from the environment, pays per call, no other setup.
- [Agent card](https://payperscreen.com/.well-known/agent-card.json):
  machine-readable service descriptor (skills, pricing, payment scheme).

## Base URL

All endpoints below are served from `https://api.payperscreen.com`
(`GET /` there is free and returns a short index).

## How payment works

No API key, no signup. A request with no payment gets back `402 Payment
Required` naming the exact price and asset (USDC) for each network you can
pay on: Base (`eip155:8453`), Polygon (`eip155:137`) or Arbitrum
(`eip155:42161`). The caller signs a gasless EIP-3009
`transferWithAuthorization` and retries with it attached; x402-aware HTTP
clients (e.g. `@x402/fetch`) do this automatically. You only need USDC on
one of those chains: no gas token, since the facilitator submits the
transaction. To see a real response first, call the free
`GET https://api.payperscreen.com/sample/screen` or `/sample/screen-address`.

## MCP server

AI agents can use PayPerScreen as tools via MCP: `npx -y payperscreen-mcp`
(tools: screen_name, screen_wallet, check_email, check_ip, plus free
lists_coverage and samples). It pays per check from the wallet in
PAYPERSCREEN_PRIVATE_KEY, with a spending cap (PAYPERSCREEN_MAX_SPEND,
default $1). Setup: https://github.com/TonytheTiger22/payperscreen/tree/main/mcp

## Endpoints (see /docs for full detail)

- `GET /screen`, $0.02: fuzzy name match across the loaded sanctions lists (OFAC, UN, EU, UK); each hit
  includes record ID, programmes, aliases, and a date-of-birth / nationality / type comparison against what you sent
- `GET /context`, $0.05: the same comparison for the top hit (now included in /screen; kept for existing integrations)
- `GET /screen-address`, $0.02: exact match against OFAC's sanctioned
  crypto-address list, plus Tether's USDT freeze list on Tron
  (on-chain, as `frozenByIssuer`) and a curated list of known hack
  proceeds (FBI notices and incident reports, as `hackFunds`)
- `GET /screen-email`, $0.02: mail-server validity, disposable-domain
  check, domain age, mailbox verification, blocklist status, typo-squat
- `GET /screen-ip`, $0.01 (beta): Spamhaus ZEN blocklist and Tor
  exit-node check, IPv4 only

## Optional

- [payperscreen.com/try](https://payperscreen.com/try): the web UI for humans,
  same API underneath, same live payments.