agency-blockchain-security-auditor

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the auditor to read protocol documentation/whitepapers and to "refer to" public third‑party sources (e.g., rekt.news, DeFiHackLabs, the SWC Registry, Trail of Bits, OpenZeppelin guides) as part of its workflow, meaning it ingests untrusted public web content that could influence actions.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly DeFi/blockchain-focused and includes concrete, actionable blockchain transaction capabilities: Foundry exploit PoC and test code that performs flash swaps, deposits/withdrawals, and token transfers; examples call Uniswap pairs, manipulate reserves, and invoke vault.withdraw(); instructions mention writing/deploying rescue contracts and executing borrows/repays. These are specific crypto/ blockchain execution patterns (transactions, swaps, contract interactions) rather than generic analysis tools, so the skill provides direct crypto-financial execution capability.