The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages authentication by reading and writing an API key to ~/.ai-hive/config.json. The implementation includes security hardening by explicitly setting the file permissions to 0600 (read/write by owner only) to prevent unauthorized access by other users on the system.
[EXTERNAL_DOWNLOADS]: The CLI script facilitates the download of generated video files and thumbnails from the AI Hive service's result URLs. These operations are restricted to the primary functional purpose of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes a user-provided prompt which is sent to the AI Hive API. While there are no explicit boundary markers or sanitization for the prompt content, the impact is limited to the generation of video assets and does not grant the data access to the underlying system or agent environment.

ai-ad-video