invision-community

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly uses the Membrane CLI to connect to Invision Community and perform actions on forum data (see SKILL.md "Use when the user wants to interact with Invision Community data" and the listed Forums/Topics/Posts), which are user-generated, third‑party community posts that the agent would fetch and process as part of its workflow, exposing it to potential indirect prompt injection.