skill-08-security-audit

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

• Third-party content exposure detected (medium risk: 0.65). The workflow ingests outsider-authored free text via runtime monitoring sources like “Rekt News” and “DeFiLlama hacks dashboard” (public web content fetched at runtime), which can include arbitrary prose that the LLM may read into its context.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly focused on DeFi/crypto operations (rug-pull checks, swap/honeypot detection, liquidity adds) and requires blockchain-related services (blockchain-rpc-server, defi-protocol-server). It explicitly describes running an eth_call simulation that "attempt [s] buy → approve → sell in single trace" to detect honeypots and evaluates token swap/liquidity behavior. This is a purpose-built crypto/DeFi tool (swaps/token interactions) rather than a generic browser or HTTP tool, so it constitutes direct financial execution capability (crypto/swaps simulation and interaction).