The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes an example shell script (examples/run.sh) that uses curl and jq to interact with the Revid API. This is a standard utility for developers to test the integration.
[EXTERNAL_DOWNLOADS]: The skill communicates with the Revid API at revid.ai to submit video rendering requests and poll for project status. This behavior is documented and necessary for the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external content from social media platforms, which introduces an attack surface for indirect prompt injection.
Ingestion points: The text and url parameters within the render_video tool calls in SKILL.md accept untrusted content from social media posts.
Boundary markers: No specific boundary delimiters are defined to separate ingested text from instructions.
Capability inventory: The skill utilizes render_video and get_project_status tools to process data and retrieve results.
Sanitization: The instructions recommend pre-cleaning text (e.g., stripping URLs and handles) to improve the quality of the generated voiceover, which provides a minor level of sanitization.

revid-tweet-to-talking-head