materials-databases

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow fetches free-form text/JSON from outsider-operated public web services (AFLOW via requests.get(...).text / .json() and Materials Project via mp-api HTTP responses), which can include arbitrary strings that may be inserted into the agent’s LLM context.