clay

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

• Third-party content exposure detected (medium risk: 0.30). In the required runtime toolflow, an agent can invoke Clay read tools (e.g., findRecord, listRecords, getTable) which fetch row/cell JSON from Clay’s API (https://api.clay.com/v3/...) based on user-provided inputs like filters, so outsider-authored free text stored in the user’s Clay tables can be ingested back into the LLM.