mcp-oauth-remote-gateway

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted metadata and responses from remote OAuth servers to configure authentication and persist tokens, creating a surface for indirect prompt injection.\n- Ingestion points: OAuth resource metadata (e.g., /.well-known/oauth-protected-resource) and token exchange responses from authorization servers.\n- Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions for remote data.\n- Capability inventory: Access to terminal (shell), execute_code (Python environment), and file system write capabilities to $HERMES_HOME/mcp-tokens/.\n- Sanitization: Implements regex-based filename sanitization for security, though the contents of the remote JSON payloads are not sanitized against injection patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 02:04 AM
Security Audit — agent-trust-hub — mcp-oauth-remote-gateway