The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest data from external websites to generate video scripts and storyboards. This creates a surface where malicious content on a captured website could attempt to influence the agent's behavior.\n
Ingestion points: extracted/visible-text.txt and extracted/asset-descriptions.md are read in references/step-3-script.md and references/step-4-storyboard.md to drive content generation.\n
Boundary markers: No specific delimiters or warnings to ignore instructions within the captured data are implemented.\n
Capability inventory: The skill possesses the ability to execute shell commands (npx, node) and write files to the project directory.\n
Sanitization: There is no evidence of filtering or sanitizing the extracted web text before it is used in prompts for downstream sub-agents.\n- [REMOTE_CODE_EXECUTION]: The workflow requires frequent execution of the hyperframes CLI tool using npx (e.g., npx hyperframes capture, npx hyperframes tts, npx hyperframes validate). This execution of remote code from the npm registry is central to the skill's utility but represents a dependency on an external package.\n- [COMMAND_EXECUTION]: Instructions direct the agent to run various shell commands, including npx utilities and local scripts such as scripts/animation-map.mjs for project validation.\n- [EXTERNAL_DOWNLOADS]: The skill references external JavaScript libraries from cdn.jsdelivr.net, including the GSAP animation engine and dotLottie player, for inclusion in the generated video compositions.\n- [DYNAMIC_EXECUTION]: The skill generates HTML/JavaScript compositions for video beats. These files are constructed at runtime by the agent based on templates and storyboard data, which are then executed within a validation environment (headless Chrome).

website-to-hyperframes