binance-agentic-wallet
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, such as blockchain metadata (token symbols and names) and third-party market analysis reports from CoinMarketCap and BNB Chain Agent Studio. While this provides an attack surface for indirect prompt injection, the skill includes explicit 'Security Policy' instructions for the agent to treat this data as untrusted, never interpret it as instructions, and ignore attempts to bypass security filters.
- [COMMAND_EXECUTION]: The skill utilizes the
bawCLI tool for wallet management. The instructions follow best-practice safety protocols by requiring the agent to poll for on-chain finality before confirming transaction success to the user, preventing misinformation about the state of user funds. - [EXTERNAL_DOWNLOADS]: The skill fetches asset eligibility lists and real-time research reports from official vendor domains (
binance.com) and established industry services (coinmarketcap.com). These network operations are strictly informative and are not used for executing remote code. - [DATA_EXFILTRATION]: The skill includes a clear policy for protecting user privacy, specifically forbidding the logging or transmission of session tokens, API keys, private keys, or seed phrases, and instructing the agent to redact sensitive fields from any displayed CLI output.
Audit Metadata