twitter-research

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

• Third-party content exposure detected (medium risk: 0.30). Outsider-authored free text from the user prompt flows into the agent’s selected tweet/advanced_search call as the query parameter (e.g., catalog.selat.ai/twitter/tweet/advanced_search?query=${query}), so the agent ingests attacker-controlled search grammar at runtime.