memory-literary-analysis

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

• Third-party content exposure detected (high risk: 0.75). The skill’s Phase 2 “Source Text Preparation” requires ingesting the full literary work’s chapter text (from a file or runtime-fetched source like Project Gutenberg or other URLs), and that readable chapter prose is then written into chapter/entity notes that the agent’s LLM uses for analysis—so if the work text comes from an outsider source, it becomes free-form outsider-authored text in the LLM context.