The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external project briefs and user feedback to influence the creative direction and the automated critique process.
Ingestion points: Project briefs, brand assets, and user notes are accessed from the project context as described in SKILL.md and references/gauntlet-loop.md.
Boundary markers: The instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) when the agent interpolates untrusted user notes into prompts for sub-agents or generation tools.
Capability inventory: The skill executes local Node.js scripts for mechanical testing and calls external image/video generation tools.
Sanitization: There is no evidence of input sanitization or validation for the text ingested from the project brief or user notes.
[COMMAND_EXECUTION]: The skill includes instructions to run Node.js test scripts located in the scripts/ directory. These tests use the node:assert module to verify the deterministic behavior of the motion and scene kernels.
[EXTERNAL_DOWNLOADS]: The skill references external URLs on higgsfield.ai for the purpose of comparing generated films against industry benchmarks. These are static references to marketing and feature pages for art direction.

recoup-internal-content-design-motion