Documentation

Everything you need to understand, integrate, and demo PagePay's x402 + Algorand payment flow — from first quote to on-chain proof.

Guides

x402 Protocol Specs

HTTP 402 headers, v2 exact-AVM scheme, payload structure, error codes, and sequence diagrams.

Algorand Settlement

Testnet configuration, ASA 10458941, GoPlausible facilitator, atomic transaction groups, and explorer proof.

API Reference

Full reference for /api/summarize, /api/compare, /api/receipt, /api/trust-score, /api/audit/verify, and /api/tools.

Integrations & AI

Pera Wallet ARC-0001, GoPlausible, Groq AI, and agent spend policy configuration.

What is PagePay?

PagePay is an autonomous, machine-to-machine pay-per-page AI processing platform built on top of standard HTTP 402 Payment Required semantics and Algorand Testnet (USDC ASA 10458941).

Unlike traditional SaaS products that rely on user signups, API keys, monthly credit card subscriptions, or manual invoicing, PagePay enables AI agents, web apps, and human clients to pay for high-value LLM processing per parsed document page in real time.

PagePay includes 5 specialized AI extraction modes, dual-document side-by-side comparison, a client-side Agent Spend Policy Guard, a cryptographic SHA-256 tamper-evident log chain, a public receipt verification service, and an Algorand agent trust score lookup API.

Core Platform Capabilities

Feature	Description	Status
5 Extraction Modes	Summary, Action Items, Key Risks, Compliance Check, Checklist	✅ Production
Multi-Doc Compare	Dual document A vs B side-by-side comparison with atomic 402 payment	✅ Production
Agent Policy Guard	Client-side spend policy caps ($/req, $/session) with pre-flight checks	✅ Production
Receipt Verification	Public `GET /api/receipt` endpoint for independent transaction verification	✅ Production
Agent Trust Score	Public `GET /api/trust-score` endpoint for Algorand address reliability (0-100)	✅ Production
SHA-256 Audit Trail	Cryptographic hash chaining for tamper-evident request/payment logs	✅ Production
Startup Log Seed	Automatic restoration of real, verifiable testnet transactions on boot	✅ Production

System Architecture

PagePay splits into four cooperating layers:

Layer	Responsibility	Technology
Client / Agent	Quote, sign exact-AVM USDC, enforce spend policy, retry	React, Pera Wallet, `@x402-avm`
API Gateway	Enforce 402, decode payment, hash audit log, fulfill AI work	TanStack Start, Nitro Server
Facilitator	Verify ARC-0001 signatures, submit atomic transaction groups	GoPlausible Hosted Service
Settlement	On-chain USDC transfer	Algorand Testnet (ASA 10458941)

Unpaid Path: Client sends POST /api/summarize or POST /api/compare → Gateway returns HTTP 402 Payment Required with base64 PAYMENT-REQUIRED quote headers.

Paid Path: Client signs exact-AVM USDC transfer → retries with PAYMENT-SIGNATURE header → Gateway verifies via GoPlausible facilitator → settles on-chain → executes Groq LLM processing → appends SHA-256 audit entry → returns 200 OK with summary/comparison + txId.

Pricing & Metering Model

PagePay uses exact pricing — the amount in the 402 quote must match the signed transaction exactly.

Unit: 1 parsed document page (500 words or 1 PDF page)
Rate: $0.01 USD per page
Asset: USDC Testnet ASA 10458941 (6 decimals)
Atomic Amount: 10000 micro-units per page (= $0.01)

Example: A 3-page document quote is $0.03 → client signs a transfer of 30000 atomic USDC units to the merchant payTo address.

Dual-document comparison combines total pages: 1 page in Doc A + 1 page in Doc B = 2 pages → $0.02 combined payment.

Quick Start for Developers & Agents

1. Run Locally

npm install
npm run dev

Open http://localhost:8080 in Chrome or Edge.

2. Machine-Readable Agent Discovery

AI agents can inspect GET /api/tools at any time to discover available routes, pricing rules, payment requirements, and extraction modes.

3. Fund Testnet Wallet

Get testnet ALGO from the Algorand Dispenser and testnet USDC from the Circle Faucet.