Privacy Layer

Privacy-first
infrastructure for Agents

Building privacy-preserving protocols for the agent economy.

Privacy First

Agent Native

Open Source

Drop-in replacement

Privacy in one line

Same API. Same simplicity. Complete privacy.

Standard - x402Exposed

import { x402 } from '@x402/sdk'

const payment = await x402.pay({
  amount: 100,
  recipient: merchant,
  // Payment history linkable
  // Browsing patterns exposed
})

Private - px402Private

import { px402 } from '@prxvt/sdk'

const payment = await px402.pay({
  amount: 100,
  recipient: merchant,
  // Unlinkable credentials
  // Zero-knowledge proofs
})

Installation

$ npm install @prxvt/sdk

Our Products

Privacy-preserving infrastructure for the agent economy

x402 Payment Privacy

Our first product. Unlinkable payments for AI agents that keep browsing patterns confidential while staying compliant.

More Coming Soon

We're building a comprehensive suite of privacy tools for AI agents. Stay tuned for announcements.

Open Source

All PRXVT protocols are open source. Built by the community, for the community.

How x402 Works

Our first product: privacy-preserving payment protocol

Anonymous Credential Issuance

Agent obtains blinded credentials from payment provider without revealing identity or destination.

Privacy-Preserving Payment

x402 challenge is met with unlinkable proof. Merchant verifies payment without learning agent history.

Selective Compliance

Regulatory requirements met through zero-knowledge proofs, revealing only what's legally required.

x402 Privacy Protocol

Cryptographic separation of payment and identity

PRXVT

x402

compliant

secure

Privacy for Every Agent Interaction

Protecting agent privacy across payments, data, and beyond

Private Payments (x402)

Agents pay for services without revealing browsing patterns or creating linkable payment trails.

Confidential Data Access

Access sensitive data and APIs while protecting query patterns and usage behavior.

Anonymous Compute

Purchase compute resources without revealing workload types or model architectures.