safe

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow ingests outsider-authored free text when it fetches Safe Transaction Service data about pending/queued multisig transactions (which can be proposed by arbitrary external users), e.g., via apiKit.getTransaction(safeTxHash) / apiKit.getPendingTransactions(safeAddress) and then uses the returned transaction fields at runtime.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill is explicitly a crypto wallet / smart-account SDK (Safe). It includes APIs and code examples to create, sign, propose, and execute on-chain transactions (ETH transfers, ERC‑20 calls, batch MultiSend), deploy Safes, enable modules that can execute transactions without owner signatures, and use Transaction Service endpoints. These are direct crypto/blockchain financial execution capabilities (signing and sending transactions, managing wallets/modules), not generic tooling.