The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow to ingest untrusted data from local design files to inform code generation, creating a surface for indirect instructions to influence the agent.
Ingestion points: The agent is instructed to read design/grounding.md, reference-anchors.md, and project-specific DESIGN.md files.
Boundary markers: The skill uses a precedence rule stating that grounded references provide specific tokens while the skill's own rules enforce discipline, but it lacks explicit "ignore instructions" delimiters for the external content.
Capability inventory: The skill generates executable React/Next.js code and provides shell commands for dependency management.
Sanitization: No sanitization or validation of the ingested markdown/text content is specified.
[COMMAND_EXECUTION]: The instructions explicitly direct the agent to output shell commands (e.g., npm install <package-name>) when it detects missing dependencies in the user's project environment.
[EXTERNAL_DOWNLOADS]: The skill requires the use of several external third-party libraries including framer-motion, lucide-react, zustand, @phosphor-icons/react, @radix-ui/react-icons, and @gsap/react to achieve its design goals.

taste-skill