The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The SKILL.md file provides installation instructions that require the use of sudo to install system-level dependencies such as libcairo2-dev, libpango1.0-dev, and texlive-latex-extra.
[COMMAND_EXECUTION]: The production pipeline involves executing various command-line tools such as manim, ffmpeg, ffprobe, and pdflatex to render and process video content.
[DYNAMIC_EXECUTION]: The skill documentation in references/rendering.md includes examples of generating and executing Python code dynamically using shell heredocs. The core rendering process also executes Python code generated for Manim scenes.
[INDIRECT_PROMPT_INJECTION]: The skill processes user concepts to generate and execute Python code, presenting an injection risk. 1. Ingestion points: User topics and technical concepts provided in the planning phase. 2. Boundary markers: No explicit delimiters or instructions are provided to the agent to separate user data from generated code logic. 3. Capability inventory: The skill possesses the ability to execute code via the manim CLI and run arbitrary system commands. 4. Sanitization: There is no documentation of validation or sanitization of input concepts before they are used to generate scene scripts or narration.

manim-video