The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the installation of external agent skills and plugins. It fetches configuration and components from well-known services such as heygen-com/hyperframes and remotion-dev/skills. However, it also instructs the user to install the buildwithhanif/claude-animation-skill from a third-party community repository via the marketplace.
[COMMAND_EXECUTION]: The skill extensively uses shell commands to perform its core functions, including ffmpeg for video encoding and subframe blending, node and npx for executing rendering scripts and installing dependencies, and python for audio processing and beat detection.
[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for the agent to ingest data from untrusted external sources, specifically by visiting product URLs to capture screenshots and analyze brand identity.
Ingestion points: External product websites processed via Playwright to extract screenshots, logos, and style information as described in SKILL.md and references/prompting-and-briefs.md.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat the retrieved web content as untrusted or to ignore embedded instructions that might be present in the target site's DOM or metadata.
Capability inventory: The agent possesses capabilities to write files to the local system, perform network operations, and execute shell commands (ffmpeg, node, python).
Sanitization: The workflow does not specify any sanitization, filtering, or validation steps for the data retrieved from external URLs before it is used to influence the code generation or animation parameters.

motion-studio