The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code snippets, diffs, and PR patches from user messages or attachments. Ingestion points: The skill instructions describe reading code from messages, attachments, or screen recordings. Boundary markers: No explicit delimiters or instructions are provided to the agent to isolate or ignore potential instructions embedded within the user-provided code. Capability inventory: The skill utilizes specialized media tools including add_layout, add_overlay, and create_clip. Sanitization: There is no automated sanitization described, though the agent is instructed to 'type it out exactly' and avoid inventing code.
[COMMAND_EXECUTION]: The skill invokes platform-specific tools to generate and manage video content. Tools: The skill utilizes add_layout, add_overlay, create_clip, get_filmstrip, and get_motion_frame. Mitigation: The ask_form tool is used to present the animation plan to the user for approval, serving as a critical safety check.

code-walkthrough