keyword-research

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly calls the external Keywords Everywhere MCP APIs in Step 3 (Get Related Keywords, Get "People Also Search For"), Step 3b (Get Keyword Data), and Step 4 (Get Domain Keywords) to ingest public keyword and competitor-domain data which the agent reads and uses to drive clustering and prioritization, so untrusted third‑party content can materially influence its actions.