solidity-audit

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

• Third-party content exposure detected (high risk: 0.85). The skill ingests outsider-authored Solidity source code from the operating user’s provided target path (e.g., git diff main --name-only or --contract <path>), and then reads it as plaintext via scan.py (sol_file.read_text() and full_content += sol.read_text()), which is fed into the agent’s LLM context during report generation/analysis.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly a blockchain/DeFi smart-contract auditing tool. It is specifically designed to analyze, generate Proof-of-Concept exploit contracts (e.g., attacker contract with receive() for reentrancy, oracle manipulation PoCs, signature replay PoCs) and to run Foundry/forge tests that execute those PoCs against targets. Step 8 explicitly instructs generating and running exploit tests that perform on-chain actions (reentrancy drains, oracle manipulation, vault inflation), which are concrete crypto/blockchain transaction executions. Although it does not call external payment gateways or a wallet API by name, its primary purpose is to create and execute on-chain exploit transactions (crypto/blockchain execution), so it grants direct financial execution authority.