Sentence Stimulus Norming

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

• Third-party content exposure detected (high risk: 0.80). The skill's workflow and references/lexical-databases-guide.md explicitly direct the user/agent to download and query public lexical databases (e.g., SUBTLEX-US, SUBTLEX-UK, English Lexicon Project, CLEARPOND) via provided URLs and use those external values to match stimuli and drive experimental decisions, which means the agent would ingest third-party web-hosted content that can materially influence its actions.