v6.5.0 · Autonomous Agent Commerce

Agent Protocol

The infrastructure layer for autonomous agent commerce. DID-based identity, agent-to-agent messaging, escrow with dispute resolution, reputation scoring, and AP2 payment mandates.

API Documentation AI Framework

DID Identity

Decentralized identifiers for agents. Register, authenticate, and manage agent identity with cryptographic verification.

A2A Messaging

Structured agent-to-agent communication with priority queuing, delivery receipts, and protocol negotiation.

Escrow

Multi-party escrow with milestone-based release, dispute resolution, voting, and automatic expiration.

Reputation

Trust scoring with decay, boosts, and penalties. Reputation leaderboards and threshold-based access control.

Agent Lifecycle

From registration to autonomous commerce, each agent goes through a verified lifecycle with granular permission scoping.

1

Register

DID creation, capability declaration

2

Verify

KYC documents, identity checks

3

Fund

Wallet creation, initial deposit

4

Connect

Protocol negotiation, capability matching

5

Transact

Payments, escrow, messaging

Authentication & Security

Multiple authentication methods with 23 granular scopes ensure agents operate within their authorized boundaries.

DID Challenge-Response

Cryptographic authentication using Decentralized Identifiers. Agents prove identity without sharing secrets.

API Key Authentication

64-character API keys for programmatic access. Per-key scoping, rotation, and revocation.

Session Management

24-hour sessions with 5-minute nonce TTL. Session listing, individual and bulk revocation.

Permission Scopes

payments.send
payments.receive
wallet.read
wallet.transfer
escrow.create
escrow.release
messaging.send
messaging.receive
reputation.read
reputation.update
compliance.read
admin.manage

23 scopes total — showing 12 most common. Full list at /agent-protocol/auth/scopes

Digital Signatures

  • Transaction signing and verification
  • End-to-end message encryption
  • Fraud detection with 16+ risk factors
  • AML screening with high-risk country list

Financial Operations

Escrow System

  • Multi-Party Escrow

    $10 to $1M range with configurable hold periods and automatic expiration (30-day default).

  • Dispute Resolution

    Voting-based resolution for escrows above $10k. Dispute filing, evidence submission, and arbitration.

  • Event-Sourced Lifecycle

    Full audit trail: created, funded, held, released, expired, cancelled, disputed, resolved.

Agent Wallets

  • Multi-Currency Support

    10+ supported currencies with per-asset fee rates (0.5%–2.5%). Fund from main accounts or external sources.

  • Transaction Limits

    Configurable daily ($10k) and per-transaction ($5k) limits. Automatic escalation when exceeded.

  • Fee Structure

    2.5% standard rate, $0.50 minimum, $100 maximum. Fee calculations recorded as events.

AP2 Payment Mandates

Google's Agent Payments Protocol v2 adds structured authorization for autonomous agent spending. Three mandate types cover every agent commerce scenario.

Cart Mandate

Human-present shopping. Agent builds a cart, user approves the total, payment executes.

  • W3C PaymentRequest format
  • Itemized cart with merchant DID
  • User approval required

Intent Mandate

Autonomous spending. Agent receives a budget and natural-language intent, acts within constraints.

  • Natural language intent description
  • Budget cap with constraint rules
  • Delegator + agent DID binding

Payment Mandate

Direct agent-to-agent payment. Fixed amount, specific payee, with payment method preferences.

  • Payee DID with fixed amount
  • Bridges to x402 and MPP rails
  • VDC-backed authorization

Reputation System

Trust scoring that evolves with every interaction. Agents build reputation through successful transactions, and face penalties for disputes or failures.

Initial Score 50/100
Decay After Inactivity 30 days
Score Components Weighted
Public Leaderboard Active

Agent KYC

Three verification tiers unlock progressively higher transaction limits and capabilities.

Basic

Identity document verification

Limited

Enhanced

+ Business registration, proof of address

Higher Limits

Full

+ Biometric verification, compliance officer review

Unlimited

Event-Sourced Architecture

Every agent action is recorded as an immutable event. 10 aggregates and 60+ domain events provide a complete audit trail for regulatory compliance.

Agent Identity

7 events

Agent Wallet

4 events

Capabilities

6 events

Transactions

10 events

Escrow

10 events

Reputation

5 events

Payments

6 events

Messaging

7 events

Security

4 events

Mandates

6 events

API Surface

Discovery

Public — no auth required

  • GET /.well-known/ap2-configuration
  • GET /agents/discover
  • GET /agents/{did}
  • GET /protocol/versions

Authentication

Public — returns session tokens

  • POST /auth/challenge
  • POST /auth/did
  • POST /auth/api-key
  • POST /auth/validate

Agent Operations

Agent-authenticated

  • POST /agents/{did}/payments
  • POST /escrow/create
  • POST /agents/{did}/messages
  • POST /agents/{did}/reputation

Related Capabilities

AI Framework

6 specialized agents, 24 MCP tools, ML anomaly detection, and Temporal workflow orchestration.

Machine Payments (MPP)

Multi-rail HTTP 402 payments. AP2 mandates bridge directly to MPP and x402 payment rails.

AI Agent Commerce

Virtuals Protocol integration with TrustCert identity, spending limits, and Pimlico enforcement.

Give Your Agents a Bank Account

Register an agent, fund its wallet, and let it transact autonomously — with escrow protection, reputation tracking, and compliance built in.

Register an Agent Protocol Documentation