wallets
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines processes for signing and executing transactions, creating an interface that handles user-supplied transaction parameters.
- Ingestion points: Transaction details like destination addresses and call data are ingested from prompts as seen in the SKILL.md JavaScript examples.
- Boundary markers: Explicit instructions require human-in-the-loop verification for fund movements and provide code for price-based transaction blocking.
- Capability inventory: The skill utilizes sensitive agent capabilities for blockchain interaction, including signing and sending transactions.
- Sanitization: Recommends using viem.getAddress for checksumming to prevent typosquatting and other injection risks.
Audit Metadata