Built for the agentic web

The trust layer for
autonomous agents

Verify AI agents, enforce spending mandates, and process x402 payments before a single transaction is accepted.

mandate — x402 payment flow
Agent SDK
import { MandateClient } from '@mandate/sdk'

const client = new MandateClient({
  apiKey: process.env.MANDATE_API_KEY,
})

const data = await client.fetchWithPayment(
  'https://api.example.com/premium/data',
  {
    agentId: 'agt_research_bot',
    mandateId: 'mnd_data_only_10usd',
  }
)
// Handles 402, policy check,
// proof generation, and retry
Seller Middleware
import { requireMandate } from '@mandate/middleware'

app.get(
  '/premium/data',
  requireMandate({
    price: '0.10',
    currency: 'USDC',
    category: 'data',
  }),
  (c) => {
    const agentId = c.get('mandateAgentId')
    return c.json({
      data: '...',
      paid_by: agentId,
    })
  }
)

Designed for agents built on

OpenAI·Anthropic·LangChain·Vercel AI SDK·LlamaIndex
The problem

Agents can act.
But should they?

When an AI agent calls a paid API, the seller has no way to know if it is authorized, who owns it, what it is allowed to spend, or whether it can actually pay.

Existing payment infrastructure assumes a human is at checkout. Existing API keys give agents unlimited access. Neither model works for autonomous agents.

×Agent has unrestricted access to your wallet
×Seller cannot verify who sent the request
×No audit trail if something goes wrong
Scoped mandate limits what the agent can spend
Seller verifies agent identity before charging
Every decision logged to an immutable audit trail
How it works

Authorized in milliseconds

01
01

Register an agent

Give your AI agent a stable identity and attach a mandate defining its spending authority, allowed merchants, and expiry.

02
02

Request is verified

When the agent calls a protected endpoint, Mandate checks its identity, mandate validity, and policy in under 300ms.

03
03

Payment is settled

A cryptographic proof is generated, the mandate budget is charged atomically, and the resource is returned.

< 300ms
Verification latency
7
Policy checks per request
100%
Audit coverage
0
Keys exposed to agents
Features

Everything you need to trust your agents

Agent Registry

Every agent gets a stable, verifiable identity. Revoke access instantly if an agent is compromised.

Scoped Mandates

Define exactly what each agent can do: max spend per transaction, total budget, allowed merchants, and expiry.

Policy Engine

Seven checks on every transaction: agent status, mandate validity, merchant allowlist, category, per-tx limit, total budget, and risk.

x402 Payments

Native support for the x402 HTTP payment protocol. Sandbox mode works without any blockchain. Production settles in USDC.

TypeScript SDK

One function call handles the entire 402 challenge, policy check, proof generation, and retry.

Audit Trail

Every verification, policy decision, and payment is logged. Queryable by agent, mandate, merchant, and time.

Start verifying agents today

Free sandbox. No credit card. Full x402 flow in under an hour.