blockrun

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

• Third-party content exposure detected (medium risk: 0.30). Source SKILL.md describes a runtime path where an outsider can supply free-text that the agent sends as a live-search query to xAI Grok Live Search (search=True/search_parameters) to ingest X/Twitter content and associated text.

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

• Direct money access detected (high risk: 1.00). The skill explicitly provisions and manages an on-chain wallet (USDC) and describes autonomous payments from that wallet to pay for services. It exposes SDK functions like setup_agent_wallet(), client.get_balance(), client.get_wallet_address(), client.get_spending(), and shows the wallet being funded via QR and used to pay per-call costs (e.g., "your wallet pays per token", "pay for it autonomously"). These are concrete crypto/blockchain wallet/payment capabilities (on-chain USDC wallet and automatic micropayments), which qualify as Direct Financial Execution.