Bankr x402 SDK - Wallet Operations
Fail
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: HIGHPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- PROMPT_INJECTION (HIGH): The skill documentation reveals a significant indirect prompt injection surface for an agent using this SDK.
- Ingestion points: Natural language prompts are passed directly into the
promptAndWait()andprompt()methods in SKILL.md. - Boundary markers: No delimiters or isolation instructions are provided to separate instructions from untrusted user data.
- Capability inventory: The SDK can execute financial transactions (e.g., 'Swap 0.1 ETH to USDC') and sign micropayments using a private key.
- Sanitization: The skill does not mention or implement any input sanitization or validation mechanisms for the prompt string.
- EXTERNAL_DOWNLOADS (MEDIUM): The skill requires the use of an external dependency from an unverified source.
- Evidence: Dependency on the
@bankr/sdkpackage referenced in SKILL.md. - Risk: The repository and organization for this SDK are not within the recognized trusted scope, making the package integrity unverifiable during static analysis.
Recommendations
- AI detected serious security threats
Audit Metadata