Integrity and Source Verification
Every payload that you receive from a payment processor carries two inherent risks: (i) Data tampering risk (ii) Impersonation risk
Prism provides strong run-time checks to eliminate both risks. This section also includes some recommended best practices to developers using the library.
| Verification Type | What It Checks | Attack Prevented |
|---|---|---|
| Integrity | Amount, currency, and transaction ID match your records | Data tampering |
| Source | Cryptographic signature using shared secrets | Impersonation, forged webhooks |
Data Tampering (Integrity Risk)
An attacker intercepting a webhook can modify the payload before it reaches your server. The attacker will be able to exploit by:
- Changing a failed payment status to "succeeded", which might deceive you to ship unpaid orders
- Modifying the amount from $100 to $1, effectively making the customer pays less than expected.
Impersonation (Source Risk)
It is possible for attackers can forge webhooks that appear to come from payment processors. This can be exploited by:
- Sending fake "payment succeeded" webhooks, which might deceive you to ship unpaid orders
- Mimicking refund notifications, to manipulate your accounting systems
Prism provides built-in verification for both risks, and it is strongly recommended to enable them and test them before using on production.
How Prism helps with Integrity and Source Verification?
Request and Response Comparison
Prism uses a FlowIntegrity trait to compare request and response data in a strongly typed fashion. The core implementation is available in backend/interfaces/src/integrity.rs:
/// Trait for integrity objects that can perform field-by-field comparison
pub trait FlowIntegrity {
/// The integrity object type for this flow
type IntegrityObject;
/// Compare request and response integrity objects
fn compare(
req_integrity_object: Self::IntegrityObject,
res_integrity_object: Self::IntegrityObject,
connector_transaction_id: Option<String>,
) -> Result<(), IntegrityCheckError>;
}
/// Trait for data types that can provide integrity objects
pub trait GetIntegrityObject<T: FlowIntegrity> {
/// Extract integrity object from response data
fn get_response_integrity_object(&self) -> Option<T::IntegrityObject>;
/// Generate integrity object from request data
fn get_request_integrity_object(&self) -> T::IntegrityObject;
}
Amount and Currency Verification
During the payment authorization step, Prism extracts amount and currency from the request and compares with the response during run-time. The core implementation is available in backend/interfaces/src/integrity.rs:
// From backend/interfaces/src/integrity.rs
impl<T: PaymentMethodDataTypes> GetIntegrityObject<AuthoriseIntegrityObject>
for PaymentsAuthorizeData<T>
{
fn get_response_integrity_object(&self) -> Option<AuthoriseIntegrityObject> {
self.integrity_object.clone()
}
fn get_request_integrity_object(&self) -> AuthoriseIntegrityObject {
AuthoriseIntegrityObject {
amount: self.minor_amount,
currency: self.currency,
}
}
}
Webhooks payloads also include amounts (might vary across payment processors). Prism verifies these match your records.
{
"event_type": "PAYMENT_INTENT_SUCCESS",
"amount": {
"minor_amount": 5999,
"currency": "USD"
},
"expected_amount": {
"minor_amount": 5999,
"currency": "USD"
}
}
If amounts mismatch, Prism flags the discrepancy clearly. In such cases you should reject the webhook and use direct server-to-server APIs to cross validate the information.
{
"error": {
"code": "AMOUNT_MISMATCH",
"message": "Webhook amount does not match expected amount",
"webhook_amount": 4999,
"expected_amount": 5999,
"currency": "USD"
}
}
Signature Verification
Typically Payment processors sign webhooks with a shared secret, to verify the payload against pre-configured secrets. An Authorize.net might use a SHA512, whereas a PPRO might use a SHA256.
Prism handles the signature verification across multiple processors.
And below are real examples from one of the connectors on how it is implemented backend/connector-integration/src/connectors/authorizedotnet.rs:
fn verify_webhook_source(
&self,
request: RequestDetails,
connector_webhook_secret: Option<ConnectorWebhookSecrets>,
) -> Result<bool, error_stack::Report<WebhookError>> {
let webhook_secret = match connector_webhook_secret {
Some(secrets) => secrets.secret,
None => return Ok(false),
};
// Extract X-ANET-Signature header (case-insensitive)
let signature_header = request
.headers
.get("X-ANET-Signature")
.or_else(|| request.headers.get("x-anet-signature"))?;
// Parse "sha512=<hex>" format
let signature_hex = match signature_header.strip_prefix("sha512=") {
Some(hex) => hex,
None => return Ok(false),
};
// Decode hex signature
let expected_signature = match hex::decode(signature_hex) {
Ok(sig) => sig,
Err(_) => return Ok(false),
};
// Compute HMAC-SHA512 of request body
use common_utils::crypto::{HmacSha512, SignMessage};
let crypto_algorithm = HmacSha512;
let computed_signature = crypto_algorithm
.sign_message(&webhook_secret, &request.body)?;
// Constant-time comparison to prevent timing attacks
Ok(computed_signature == expected_signature)
}
If verification fails, Prism flags the discrepancy very clearly.
{
"error": {
"code": "SIGNATURE_VERIFICATION_FAILED",
"message": "Webhook signature does not match payload",
"connector": "stripe",
"suggestion": "Check your webhook secret is correct and the payload was not modified"
}
}
Recommendations for Developers
Verify the Transaction ID and Amount in your system
It is strongly recommended to track transaction IDs across the lifecycle. When a webhook or API response arrives, check the following parameters in your application database, before updating them.
| Check | If check fails? |
|---|---|
| ID exists in system | Reject unknown transaction IDs |
| Status transition valid | Reject invalid state changes (e.g., SUCCEEDED → PENDING) |
| Amount match | Reject payload to prevent amount tampering |
| Currency match | Reject payload to prevent currency tampering |
Always Configure the secrets while enabling a processor
Some payment processors may have the secrets as optional configuration/ implementation. Always, generate new secret in processor dashboard and update the Prism configuration accordingly.
An example configuration for Stripe as below.
const { PaymentClient } = require('hyperswitch-prism');
// Webhook secrets are configured per-connector in connectorConfig
// Note: Webhook verification is done at the application level
// by comparing signatures using the webhook secret
const config = {
connectorConfig: {
stripe: {
apiKey: { value: process.env.STRIPE_API_KEY }
}
}
};
const paymentClient = new PaymentClient(config);
Next Steps
- Error Handling — Handle verification failures in your application