The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's documentation instructs the agent to execute a local shell command: node scripts/render-all.mjs --still --scale=1 <ShotId>. The <ShotId> parameter is a dynamic value derived from the user's request or the generated component's name. Although the skill specifies a PascalCase naming convention, an adversarial user could provide a description that results in a name containing shell metacharacters, potentially leading to arbitrary command execution on the host machine.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it consumes untrusted data (user-provided video and shot descriptions) to generate executable TSX code and to form arguments for shell commands.
Ingestion points: User descriptions of scenes or animations entering the agent context via SKILL.md.
Boundary markers: No delimiters or protective warnings are provided to prevent the agent from obeying instructions embedded within the user data.
Capability inventory: The skill has the capability to write files and execute shell commands via the node process.
Sanitization: No programmatic sanitization or validation of the user input is implemented prior to its use in the shell command or component generation.
[EXTERNAL_DOWNLOADS]: The skill relies on external Node.js dependencies and a local script. While packages like remotion and chroma-js are well-known, the skill references an external script scripts/render-all.mjs that is not part of the skill's own file set, meaning its behavior is unverified.

vidtsx-2d-generator