The Agent Skills Directory

[SAFE]: A detailed audit of all 17 files, including the setup script and reference documentation, revealed no malicious patterns or security risks.
[PROMPT_INJECTION]: The skill contains pedagogical instructions for video creation and does not include any attempts to override agent behavior, bypass safety filters, or extract system prompts.
[DATA_EXFILTRATION]: No access to sensitive file paths (e.g., .env, .ssh) or patterns for unauthorized data exfiltration were found. The skill focuses on generating local animation assets.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts. The scripts/setup.sh utility is used solely for checking the presence and versioning of local prerequisites like Python, Manim, and LaTeX.
[COMMAND_EXECUTION]: Shell command usage is limited to standard Manim rendering (manim) and ffmpeg stitching operations required for the animation pipeline.
[DYNAMIC_EXECUTION]: The skill involves the generation of Python animation code, which is the primary purpose of the Manim library. The instructions follow established development patterns without using unsafe practices like hidden eval() or unauthorized library injection.

manim-video