igrantio-api-holder

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by instructing the agent to ingest and process external URIs which may contain untrusted data.\n
  • Ingestion points: The skill handles untrusted data through openid-credential-offer:// and openid4vp:// URIs via the /receive endpoints in SKILL.md.\n
  • Boundary markers: The instructions lack specific delimiters or guidelines for the agent to ignore potentially malicious instructions embedded within these external URIs.\n
  • Capability inventory: The skill enables network operations (GET, POST, PUT, DELETE) to the iGrant.io API based on the processed inputs.\n
  • Sanitization: While the skill mentions the backend API filters metadata, there are no explicit agent-side sanitization or validation steps for the raw URI inputs documented.\n- [SAFE]: The skill provides best-practice guidance for authentication, specifying that API keys must be sent via the Authorization header and should remain on the server rather than the client browser.\n- [SAFE]: All network endpoints and documentation links refer to official vendor domains (igrant.io), which are appropriate and expected for the skill's stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 03:58 PM
Security Audit — agent-trust-hub — igrantio-api-holder