1claw

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime setup explicitly runs remote code via npx (e.g. "npx -y @1claw/mcp@0.32.2"), which fetches and executes the @1claw/mcp package from the npm registry (https://www.npmjs.com/package/@1claw/mcp) and is a required runtime dependency, so it executes external code that controls the agent's tool runtime.

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

• Direct money access detected (high risk: 1.00). The skill explicitly exposes crypto and banking execution capabilities: multi-chain signing keys (Ethereum, Bitcoin, Solana, XRP, Cardano, Tron), transaction intent tools (submit_transaction, sign_transaction, simulate_transaction), Bankr key vending (lease_bankr_key, storing bk_ keys, CLI/SDK operations to PUT/Retrieve Bankr API keys), and treasury multisig proposal/signing (treasury_propose, treasury_sign_proposal with auto-execute). These are specific, purpose-built functions for signing/broadcasting on-chain transactions and interacting with a wallet/payment provider (Bankr). Although guardrails and human-confirmation are required, the skill’s primary and explicit definitions are to perform financial transactions and key management, so it grants Direct Financial Execution Authority.