vos-remix

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly tells the agent to "use" a remix grant if it appears in the user's prompt (used as an Authorization: Bearer token), which requires the model to read and insert a secret value verbatim into requests/commands — an exfiltration risk despite the "never print them" note.

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow fetches and reads public vos gallery/program content (e.g., via vos fetch https://vos.so/vos/<id> / GET /api/vos/{id} and /api/vos/{id}/config), which can contain outsider-authored free text like declared JS function strings and user-edited notes that the agent then edits/validates/pushes.