omni-gateway-dump

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md Step 1/2 instructs the runtime to unzip and parse user-supplied dump ZIP contents (JSON files like api-instances.json, policy-bindings.json, etc.) into LLM-readable tables, which are outsider-authored free text when the dump originates outside the operating user’s chosen inputs.