gmfarcaster

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

• Third-party content exposure detected (medium risk: 0.30). At runtime, the user-provided question string is POSTed as JSON to the Warpee Knowledge API endpoint (resp = session.post(API_URL, json={"query": query}...)), and the skill then ingests and prints the API’s answer and citations from the response (data = resp.json()), meaning outsider-authored free text flows into a system that reads/writes provider-returned content that may include instructions.

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

• Direct money access detected (high risk: 1.00). The skill explicitly performs on-chain payments: it requires a wallet private key, signs x402/EIP-3009 USDC payments locally, and will spend real USDC per query. That is direct crypto payment execution (wallet signing/sending), so it grants Direct Financial Execution authority.