The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its handling of user-provided content.
Ingestion points: User scripts (口播稿) are ingested in the first phase of the pipeline for audit and processing.
Boundary markers: There are no explicit markers or instructions to isolate user input from command logic during synthesis or code generation.
Capability inventory: The skill performs file writing operations (HTML, JSON) and executes shell commands for speech synthesis and directory management.
Sanitization: The workflow lacks instructions for sanitizing or escaping user-provided text before it is inserted into the mmx speech synthesize shell command or the final JavaScript-based animation logic.
[EXTERNAL_DOWNLOADS]: The skill utilizes external software and libraries to achieve its functionality.
It recommends the installation of the mmx-cli package via npm to support text-to-speech synthesis.
It fetches the GSAP (GreenSock Animation Platform) library from the well-known CDNJS service to power the generated animations.
It provides an example for fetching audio from the OpenAI TTS API via curl.
[COMMAND_EXECUTION]: The skill's implementation involves the use of local shell commands for automation.
It executes the mmx CLI tool to synthesize audio segments from extracted text.
It uses standard shell utilities like mkdir and for loops to manage the file environment during Phase 6B.

jacky-motion