The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external video sources to generate executable React code, presenting a risk of indirect instructions being interpreted by the agent during analysis.
Ingestion points: Video downloads and frame extraction via python scripts/download_video.py and ffmpeg as described in SKILL.md.
Boundary markers: There are no explicit instructions to the agent to ignore or delimit instructions found within the visual content of the video frames.
Capability inventory: The skill has the ability to execute shell commands (ffmpeg, npm start) and write to the local file system (src/animations/).
Sanitization: The skill does not describe any sanitization or validation process for the text or patterns identified during video analysis before they are interpolated into code.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using placeholders for user-supplied URLs and file paths.
Evidence: The usage of python scripts/download_video.py "VIDEO_URL" -f in the Step 1 documentation.
Evidence: The usage of ffmpeg -i path/to/video.mp4 with a variable file path in the Alternative Step 1 documentation.
[DYNAMIC_EXECUTION]: The skill is primarily designed to generate and then execute new code based on runtime observations.
Evidence: Steps 4 through 7 in SKILL.md detail the process of generating new React components and running them using npm start.

text-animation-copier