mastodon

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

• Third-party content exposure detected (high risk: 0.75). The skill calls Mastodon’s REST API at runtime using the user’s injected token, and the LLM ingests the returned JSON (including verify_credentials fields like note/source.text and status text) which is outsider-authored content from the user’s Mastodon account/instance, not authored by the operating user in the prompt.