1inch-swap
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill handles structured data such as quotes and transaction calldata returned by the 1inch swap tool.
- Ingestion points: Data enters the agent context through the output of the swap MCP tool (SKILL.md, references/QUOTE.md).
- Boundary markers: The instructions guide the agent to present specific transaction details, approval reasons, and cost estimates to the user before proceeding with any signature or broadcast (references/QUOTE.md, references/CLASSIC.md).
- Capability inventory: The skill facilitates transaction broadcasting via the product_api tool and provides instructions for local signing using tools like cast, web3.py, and viem (references/CLASSIC.md, references/FUSION.md).
- Sanitization: The skill strongly recommends using WalletConnect, which ensures that transaction signing occurs within the user's wallet application rather than the agent's environment, mitigating potential credential exposure (SKILL.md).
- [SAFE]: The skill documentation and referenced implementation guides follow industry standards for decentralized finance (DeFi) interactions and do not contain malicious patterns, obfuscation, or unauthorized access attempts.
Audit Metadata