solidity-auditor

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches external web content — e.g., Turn 1 calls curl against https://raw.githubusercontent.com/... and assets/docs/README states that files listing one URL per line "will be fetched and read automatically" — and those fetched, user-provided/public URLs are ingested and used to inform the audit agents, so untrusted third‑party content can directly influence tool decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The eval runner clones benchmark repositories at runtime (e.g., git clone https://github.com/sherlock-audit/2025-05-dodo-cross-chain-dex per evals/runner.md) and then injects those fetched source files into agent bundles, which directly control the model context/prompts.