openfort-backend-wallets
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a high-privilege attack surface as it processes external transaction data and messages that could contain malicious instructions.
- Ingestion points: External data processed via
sendTransaction,signMessage,signTypedData, and webhook payloads received from the Openfort platform (SKILL.md). - Boundary markers: Not explicitly enforced in the instructional prompt; requires the agent to interpret transaction intents which may be influenced by external content.
- Capability inventory: Significant financial and cryptographic capabilities, including initiating blockchain transfers, signing arbitrary data, and importing/exporting private keys across EVM and Solana (evm-wallets.md, solana-wallets.md).
- Sanitization: While the SDK provides Zod validation schemas for API inputs (policy-engine.md), the high-level intent interpretation remains a vulnerability if the agent acts on unsanitized data from the web or user input.
- [COMMAND_EXECUTION]: The skill allows the agent to execute high-impact blockchain commands programmatically. This includes sending transactions and signing data server-side without user approval, which represents a high-privilege capability that could be misused if the agent's logic is compromised.
Audit Metadata