binance-agentic-wallet
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill interfaces with the local system through the
bawcommand-line tool to perform wallet operations. All commands are documented with specific parameters and use the--jsonflag for consistent parsing. - [EXTERNAL_DOWNLOADS]: The skill manages its execution environment by checking for updates and installing the official
@binance/agentic-walletpackage from the npm registry. These operations target the vendor's own verified infrastructure. - [PROMPT_INJECTION]: The security policy includes explicit instructions to guard against indirect prompt injection by treating on-chain data, such as token names and symbols, as untrusted input that must not be interpreted as instructions.
- [DATA_EXFILTRATION]: There is a strong emphasis on credential protection. The skill instructions strictly forbid the logging, display, or request of sensitive information like private keys, seed phrases, or session tokens.
- [SAFE]: The skill implements a 'fail-closed' security model, requiring human confirmation for all state-changing operations and utilizing a separate token audit skill to verify assets before trading.
Audit Metadata