oauth2-oidc-implementer

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines authentication logic that ingests data from external Identity Providers (IdPs), creating a potential surface for indirect prompt injection.
  • Ingestion points: Data is received from token exchange and user info endpoints within SKILL.md.
  • Boundary markers: The implementation includes the use of unique state parameters and PKCE (Proof Key for Code Exchange) to protect the flow integrity.
  • Capability inventory: The code utilizes network requests via fetch and manages session state using cookies.
  • Sanitization: It correctly utilizes the jose library to cryptographically verify signatures and claims of ID tokens and session JWTs before processing them.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill initiates network connections to external domains for authentication purposes.
  • Evidence: Requests are directed to identity provider endpoints in lib/auth/providers.ts.
  • Context: The targeted domains (Google, GitHub, Microsoft) are well-known services. The skill adheres to security best practices by using environment variables for sensitive client secrets rather than hardcoding them.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 09:31 AM
Security Audit — agent-trust-hub — oauth2-oidc-implementer