iblai-vibe-memory

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill explicitly describes admin-scoped UI and API usage that lets administrators and RBAC-granted users read, edit, delete, and toggle other users' memories and personalization settings, which weakens user privacy/security boundaries without clear malicious intent.

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow behind the /iblai-vibe-memory tenant Memory surface (via MemoryAdminTab inside <Account targetTab="memory">) reads administrator-exposed tenant memory text (e.g., global memories’ stored “text” shown/managed in the Global tab and per-agent memories in the Agent tab) that an outsider can author through the normal memory-creation path (e.g., via chats/integration or prior tenant activity), so the agent/LLM runtime could ingest free text from that managed memory content.