iblai-vibe-agent-sandbox

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

• Third-party content exposure detected (high risk: 0.70). The required runtime workflow involves pushing/reading agent prompt markdown (Identity/Soul/User Context/Tools/Agents/Bootstrap/Heartbeat/Memory) from the connected OpenClaw sandbox via the SDK’s config/push APIs, so outsider-authored free text from the sandbox (i.e., not authored by the operating user) can be ingested into the agent’s LLM context.