The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently invokes shell commands via a companion script (artifact.mjs) located in a sibling directory relative to the skill. These commands perform critical tasks such as voice synthesis, timing alignment, and video exporting. It also supports a custom command flag that allows for the construction and execution of complex bash strings for Text-to-Speech processing using system tools.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from multiple sources to generate narration and executable scene code. This includes user-provided articles, reports, external websites for fact-checking, and user comments. The process of transforming this external data into code for the film engine creates an attack surface for indirect prompt injection.
Ingestion points: User-provided materials (briefs, reports, datasets), external URLs accessed for fact-checking, and the local comments.json file.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the processed data.
Capability inventory: The skill possesses significant capabilities, including local file system access, shell command execution (node, ffmpeg, curl), and browser control (Chrome/Chromium).
Sanitization: There are no documented procedures for sanitizing or escaping content from external sources before it is interpolated into the agent's prompts or generated scripts.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to well-known services to support its core functionality. This includes fetching voice metadata from ElevenLabs (api.elevenlabs.io) and downloading fonts from Google. These references are consistent with the skill's stated purpose and target established service domains.

motion-director