interminai

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime LLM context exposure comes from interminai output/wait reading the PTY screen, which contains arbitrary free-form text produced by the executed outsider CLI (e.g., a subagent like cursor-agent), and then returning it as readable screen text in the JSON response that the supervising agent would ingest.