The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a system for generating executable code (GSAP, Framer Motion, Remotion) and styles (CSS) by interpolating user-provided specifications into templates. This creates a surface where malicious input could influence the generated output.
Ingestion points: The skill ingests AnimationSpec objects, specifically fields like name, recipe, and principles, which are then mapped to template variables such as {{ANIMATION_NAME}}, {{RECIPE_REF}}, and {{PRINCIPLES}} as documented in references/template-variables.md.
Boundary markers: While the template engine uses {{ }} delimiters, there are no instructions or mechanisms mentioned to sanitize these inputs or to prevent them from breaking out of the intended string literals within the generated code.
Capability inventory: The skill's primary purpose is code generation for platforms that execute JavaScript (GSAP, Framer Motion, Remotion). This allows for a path from untrusted input to executable code if the orchestrator does not perform external validation.
Sanitization: No explicit sanitization, escaping, or schema validation rules are provided in the skill documentation to handle potentially malicious strings in the input metadata.

code-generation