The Agent Skills Directory

[SAFE]: The skill's operation is consistent with its stated purpose of media processing and analysis. No malicious patterns such as credential theft, unauthorized persistence, or privilege escalation were detected.\n- [COMMAND_EXECUTION]: The skill utilizes Python and shell commands to invoke the VideoDB SDK. These operations are limited to the intended functionality of the skill, such as media ingestion and search, using predefined documentation templates.\n- [EXTERNAL_DOWNLOADS]: The skill recommends the installation of the official videodb and python-dotenv packages from standard registries. These are the core dependencies required for the service integration.\n- [PROMPT_INJECTION]: The skill contains workflows that process external media data through LLMs for summarization. While this presents an inherent surface for indirect prompt injection, there are no instructions present that attempt to bypass safety guidelines or override agent behavior.\n- [DATA_EXFILTRATION]: Media data and API requests are sent to the official VideoDB vendor domains at videodb.io. The skill avoids accessing sensitive local files unrelated to its media processing tasks.

videodb