wallet

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

• Third-party content exposure detected (high risk: 0.90). The skill's scripts/abi.ts explicitly fetches ABIs from public third-party services (Sourcify at https://sourcify.dev, Etherscan via its API, and an optional ABI proxy via ABI_PROXY_URL) and SKILL.md requires using list-functions/loadAbi to discover and act on those ABIs, meaning untrusted, user-generated contract metadata from the open web is read and can materially influence which functions are called and how transactions are constructed.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The deno.json import specifiers (e.g., "npm:viem@^2" and "jsr:@std/cli@^1") are fetched by Deno at runtime and execute remote code as required dependencies for the skill, presenting a runtime remote-code dependency risk.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly a multi-chain EVM wallet whose primary purpose is on-chain financial operations. It includes commands and workflows to send on-chain transactions, approve ERC‑20 allowances, move tokens, sign EIP‑191/EIP‑712 messages, generate and store a private key, and check/modify balances. Key explicit capabilities: the send command for contract writes (with --approve for token approvals), mandatory allowance checks and two‑tx approve+call flows, private-key-backed signing (PRIVATE_KEY, generate-wallet), and guidance that write operations "send transactions directly — the user pays gas." It even provides an exception where a SyncTx workflow can skip user confirmation, enabling automated execution. These are specific crypto/blockchain financial execution functions (wallet, signing, sending transactions, token approvals), not generic tooling. Therefore this skill grants direct financial execution authority.