ov-resources

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

• Third-party content exposure detected (medium risk: 0.30). The ov-resources skill’s runtime workflow can ingest outsider-authored free text via ov add-resource when the user supplies a URL (or Git repo) and then read/search the resulting content in viking://resources/ (e.g., ov read, ov find/search, ov grep), meaning queued/watched ingestion of user-provided content is possible.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill explicitly directs the agent to run destructive filesystem operations (rm --recursive, write) and to perform import/backup/restore commands that “require ROOT or ADMIN permissions”, which encourages privileged actions that can modify the host/system state.