igrantio-holder-backend

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a backend proxy architecture that ingests external data (credential offers, presentation requests) to interact with the iGrant.io OWS API.
  • Ingestion points: Untrusted request bodies in references/src/proxy.ts and query parameters in references/src/notificationsSse.ts are processed by the server.
  • Boundary markers: The skill implements a robust boundary in references/src/server.ts via the HOLDER_PERMITTED_PATHS allow-list, which uses strict string matching and negative-lookahead Regular Expressions to enforce least privilege and prevent access to non-holder endpoints (e.g., blocking issuer/verifier paths).
  • Capability inventory: The service has the capability to perform authenticated network requests to the configured owsBaseUrl using tenant-specific API keys.
  • Sanitization: The implementation uses standard Express body parsing and forwards binary data to the upstream service without interpreting the contents, relying on the upstream API for payload validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 11:41 PM