search-term-miner

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

• Third-party content exposure detected (high risk: 0.75). The skill’s REQUIRED workflow ingests the user-provided “exported search-terms report” (CSV/paste/path) as untrusted text at runtime, and that export is outsider-authored free-form content (e.g., downloaded/pasted report text from outside the agent’s own code/workspace) that can contain embedded instructions, which the LLM may read while extracting queries/cost/conversions.