chainlink-confidential-ai-attester-skill

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

• Third-party content exposure detected (high risk: 0.75). The skill’s POST /v1/inference supports an outsider-controlled resources[].url field; at runtime the server fetches that URL server-side and the fetched page/body text is ingested into the enclave/LLM context, enabling indirect prompt injection from public web content.