Stripe x402 < 2 minutes 1 line of middleware

Stripe Machine Payments + Observer Protocol

Stripe handles the payment. Observer Protocol answers the question Stripe cannot: should you trust the agent making it?

Add one middleware line to your x402 seller and every incoming machine payment gets a reputation check — trust tier, payment history, risk signals — before you process it. No API key required to start.

💳

Agent sends x402 payment to your Stripe endpoint

🔍

OP middleware queries agent reputation before you accept

✓

Trust tier, score, and suggested action — before settlement

┌─────────────────────────────────────┐

│ OBSERVER PROTOCOL LAYER │ ← "Can I trust this agent?"

└──────────────────┬──────────────────┘

│

┌──────────────────▼──────────────────┐

│ STRIPE MACHINE PAYMENTS │ ← "Process this payment"

└──────────────────┬──────────────────┘

│

┌──────────────────▼──────────────────┐

│ Base (USDC) · Solana · Tempo │ ← Payment rails

└─────────────────────────────────────┘

For Sellers

Install the SDK and add one middleware line

npm install @observer-protocol/sdk

const { ObserverClient } = require('@observer-protocol/sdk');
const observer = new ObserverClient();

// Add before your x402 payment handler — that's it
app.use('/api/*', observer.reputationMiddleware({ fallback: 'review' }));

// Reputation is now available on every request:
// req.observerReputation.trust_tier   → "new" | "active" | "established" | "trusted"
// req.observerReputation.reputation_score → 0.0 – 1.0
// req.observerReputation.suggested_action → "accept" | "review" | "reject"
        

Gate access by trust tier

Use the reputation response to apply different rate limits, pricing, or access levels per agent — automatically, before the payment settles.

app.post('/api/data', (req, res) => {
  const { trust_tier, suggested_action } = req.observerReputation || {};

  if (suggested_action === 'reject') {
    return res.status(403).json({ error: 'Agent reputation: rejected' });
  }

  // Rate limit by trust tier
  const rateLimit = {
    trusted:     10000,   // req/day
    established: 5000,
    active:      1000,
    new:         100,
    unknown:     10,
  }[trust_tier || 'unknown'];

  res.json({ data: '...', rate_limit: rateLimit });
});
        

For Agents

Reputation lookup by agent_id

Stripe identifies agents by their USDC wallet address. After a one-time registration on Observer Protocol (the agent submits its public key + metadata, receives an agent_id), sellers query reputation by that agent_id — the agent passes it via the X-Observer-Agent-ID header on every request. The middleware in step 1 reads the header and does the lookup automatically. The @observer-protocol/sdk client wraps this same call.

# Query by agent_id (returned at registration; passed via X-Observer-Agent-ID header)
curl https://api.observerprotocol.org/api/v1/agents/maxi-0001/profile

# Returns (current AT-ARS schema):
{
  "agent_id": "maxi-0001",
  "agent_name": "Maxi",
  "did": "did:web:observerprotocol.org:agents:maxi-0001",
  "verified": true,
  "trust_score": 58.0,
  "trust_data": {
    "receipt_count": 12,
    "unique_counterparties": 5,
    "last_activity": "2026-05-07T18:42:14Z",
    "components": { "receipt_score": 68, "counterparty_score": 75, "org_score": 100, "recency_score": 90, "volume_score": 45 }
  }
}
        

Direct wallet-address → reputation lookup is on the roadmap (issue open on the mpp-integration repo). Until then, register once and use agent_id everywhere.

Trust Tiers (AT-ARS Spec §6)

AT-ARS scores 0–100, weighting verified receipts (25%), counterparty diversity (20%), org affiliation (20%), recency (15%), and volume (15%). The score band is what the SDK exposes as trust_tier.

Tier	Score Range	Suggested Action
unknown	no profile in OP	review
untrusted	0–39	review
developing	40–59	review (lenient: accept)
established	60–79	accept
trusted	80–100	accept (premium)

Chargeback-resistant settlement (composition with WDK)

Reputation lookup answers "should I accept this payment?" For high-value flows that need "this payment is defensible against later disputes", compose with @observer-protocol/wdk-protocol-trust. The trust module handles bilateral identity verification (sender + recipient both prove DIDs before settlement) and ERC-8004 chain anchoring of the receipt. Together they make agentic payments chargeback-resistant by construction.

Live demo: observerprotocol.org/chargeback-prevention/wdk walks the full Mercado Libre × USDT-on-x402 flow — soft-reject → magic-link authorization → bilateral handshake → cryptographically attested receipt.

Agentic Terminal

See real verified transactions in the live feed. Reputation queries working now.

Open Agentic Terminal →

Full SDK

Node.js + Python middleware. Express, FastAPI, and raw fetch examples included.

Get the SDK →

Verification Policy

Platform-required, opt-in, or mandatory. Full policy architecture for payment processors.

Read Architecture →