mailerlite

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

• Third-party content exposure detected (high risk: 0.80). The skill's workflow (see "Popular actions" like get-campaign, list-subscribers and the "Proxy requests" section) instructs the agent to fetch and ingest data from the Mailerlite API via Membrane (e.g., membrane request CONNECTION_ID /path/to/endpoint), which can return untrusted, user-generated content (campaign bodies, landing pages, subscriber data) that the agent would read and could influence subsequent actions.