goat
Warn
Audited by Snyk on Aug 4, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). At runtime, the GOAT agent examples use the Vercel AI SDK (via
generateText) and pass the user’s free-textprompt/inputdirectly into the LLM along with onchaintools, so outsider-authored text can drive tool calls.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 1.00). The SKILL.md documents an on-chain agent framework that explicitly provides wallet providers, transaction-signing wallet clients, and protocol plugins for token transfers, DEX swaps, bridging, and custodial smart wallets (e.g., sendETH, sendSOL, erc20, uniswap, jupiter, debridge, wallet-viem, wallet-solana, wallet-crossmint). It includes examples that create wallet clients with private keys, call getOnChainTools with send/send-swap plugins, and prompt the agent to "Send 5 USDC..." or "Swap 1 SOL for USDC", and defines wallet client interfaces with sendTransaction methods — all of which are direct financial execution capabilities.
Issues (2)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W009
MEDIUMDirect money access capability detected (payment gateways, crypto, banking).
Audit Metadata