skills/vtex/skills/payment-async-flow/Gen Agent Trust Hub

payment-async-flow

Pass

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents how to handle external webhooks and browser redirects, which serve as ingestion points for untrusted data from third-party payment providers.
  • Ingestion points: Handlers handleAcquirerWebhook and callbackHandler process data from external request bodies and query parameters.
  • Boundary markers: The skill provides logic to map variable external statuses into a fixed set of approved internal values.
  • Capability inventory: The logic utilizes fetch for network notifications and vbase for transaction storage.
  • Sanitization: The instructions require preserving the X-VTEX-signature query parameter for cryptographic validation of the callback source.
  • [DATA_EXFILTRATION]: The skill implements network operations to communicate payment status updates to the platform. These requests target legitimate platform-owned domains and use authenticated headers derived from environment variables rather than hardcoded secrets.
  • [EXTERNAL_DOWNLOADS]: The skill identifies a dependency on the @vtex/payment-provider package, which is a verified platform library from a known organization.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 26, 2026, 08:40 PM