The Agent Skills Directory

[DATA_EXFILTRATION]: The skill transmits user-supplied TikTok URLs to https://www.creatok.ai/api/open/skills/analyze to perform video analysis. This operation is authenticated using the CREATOK_API_KEY environment variable. While this is the intended functionality of the skill, it involves sending data to an external service.
[PROMPT_INJECTION]: The workflow ingests and processes external data (transcripts and visual analysis) which is considered untrusted as it originates from third-party TikTok videos. This creates a surface for indirect prompt injection.
Ingestion points: Data is fetched in lib/analyze-video.js and written to outputs/recreate_source.json, which the model is then instructed to read in SKILL.md.
Boundary markers: None observed; the instructions do not mandate the use of delimiters or 'ignore' instructions when the model parses the video context.
Capability inventory: The skill has the ability to perform network requests via fetch and write files to the local system using fs.writeFileSync.
Sanitization: No sanitization or validation of the fetched transcript or analysis content is performed before it is presented to the AI model.
[COMMAND_EXECUTION]: The skill includes a Node.js script scripts/run.js designed to be executed via the command line. It handles arguments such as --tiktok_url and --run_id to orchestrate the video recreation process.

creatok-recreate-video