multi-pattern

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

• Third-party content exposure detected (high risk: 0.70). The required workflow references runtime lookups (“exa.search”, “perplexity”, and “crustdata for LinkedIn signal”), which commonly involve fetching and then ingesting public web/free-form text into the agent’s LLM context, but the exact ingestion steps aren’t shown in the provided files.