pi-coding-agent

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

• Third-party content exposure detected (medium risk: 0.30). Pi’s RPC mode ingests arbitrary outsider-authored free text via pi --mode rpc where a client can send { "type":"prompt", ... "message": ... } and that message is forwarded into the LLM context (agent-core converts messages with role "user" into LLM input).