pair-trading

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

• Third-party content exposure detected (high risk: 0.70). Yes—at runtime the required workflow fetches public web content from the OKX API (requests.get(...) → resp.json()["data"] → converted into a pandas DataFrame), so outsider-authored free-form text/JSON from an external party is ingested into the LLM context via the program’s data path.