The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill automates the installation of the 'uv' package manager by piping a remote installation script to the shell. This is a standard installation method for this tool from a well-known provider.\n
Evidence: The command curl -LsSf https://astral.sh/uv/install.sh | sh in SKILL.md.\n- [EXTERNAL_DOWNLOADS]: The skill downloads several established software dependencies, including the Puppeteer browser automation library, the rembg background removal tool, and the Codex CLI from official registries.\n
Evidence: npm i puppeteer@24, uvx --from \"rembg[cpu,cli]\", and npm i -g @openai/codex in SKILL.md and README.md.\n- [COMMAND_EXECUTION]: The skill uses shell commands to coordinate media processing tasks using tools like ffmpeg, Node.js, and Codex as part of its core functionality.\n
Evidence: allowed-tools in SKILL.md lists Bash(ffmpeg *), Bash(node *), and Bash(codex *).\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied animation descriptions which are used to generate prompts for an image generation model, creating a surface for indirect instructions.\n
Ingestion points: The <animation idea> argument in SKILL.md.\n
Boundary markers: The skill uses a structured prompt template defined in references/look.md to wrap the user's input.\n
Capability inventory: The agent has access to file writing and media processing tools like ffmpeg and rembg.\n
Sanitization: The instructions direct the agent to manually inspect generated images for unexpected content like logos or text.

vox