The Agent Skills Directory

Potential for Indirect Prompt Injection: The skill is designed to ingest and process user-provided content, such as product requirements, copy, and navigation labels, to generate design briefs and implementation plans. While this is necessary for its operation, it creates a surface where external data is interpolated into agent prompts.
Ingestion points: User-supplied requirements, copy, and labels (documented in SKILL.md and references/imagegen-website-concepts.md).
Boundary markers: The instructions do not explicitly define delimiters or "ignore" instructions for the interpolated user content.
Capability inventory: The skill utilizes browser tools, image generation, and file system operations for application implementation.
Sanitization: No specific sanitization or validation logic is defined for the user-supplied text.
Dependency Considerations: The skill defaults to using common industry frameworks and libraries for development. Users should ensure that implementation follows standard security practices for managing these dependencies.
Packages: react, vite, and playwright are mentioned as default tools for implementation and verification.
Safe Credential Management: The skill incorporates best practices for secret management by instructing the agent to use official credential flows (such as openai-platform-api-key) instead of manual environment variable setup or hardcoded placeholders.
Verification Workflow: The skill implements a robust design review process using browser snapshots and image comparison to ensure fidelity, which serves as a manual checkpoint for the developer during the implementation phase.

frontend-app-builder