The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local project files and external attachments, creating a surface for indirect instructions.
Ingestion points: Accesses .claude/design-tokens.json, ~/.claude/design-systems/, codebase tokens, and external attachments such as Figma files or PRDs.
Boundary markers: The skill does not provide specific delimiters or instructions to the agent to differentiate between project data and potentially embedded malicious instructions.
Capability inventory: Possesses capabilities to write files (Write), modify content (Edit), execute shell commands (Bash), and interact with browser sessions (Chrome DevTools), which could be misused if malicious data is processed.
Sanitization: No explicit sanitization or validation logic is present for data ingested from the local environment or external sources.
[DYNAMIC_EXECUTION]: The skill generates and executes React/JSX code at runtime using Babel standalone within the browser environment. The animation logic is dynamically composed into an HTML artifact and rendered without a build step, relying on client-side compilation and execution.
[EXTERNAL_DOWNLOADS]: The skill incorporates an external animation library from a well-known content delivery service. Evidence: Includes a script tag for https://unpkg.com/popmotion@11.0.5/dist/popmotion.min.js in generated HTML artifacts.

animated-video