The Agent Skills Directory

[SAFE]: No security issues detected. The skill focuses on guiding an AI to create well-designed, standalone HTML files without requiring external network access or privileged operations. It includes explicit instructions to avoid external dependencies by default, which minimizes the risk of supply chain attacks or data exfiltration in generated artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by design, as it instructs the agent to read and follow context from repository files (e.g., AGENTS.md, CLAUDE.md, design system documentation) and user-supplied briefs to shape the generated HTML artifact. Evidence: 1. Ingestion points: Workspace files including AGENTS.md, CLAUDE.md, design-system documentation, design tokens, existing components, and user-provided briefs. 2. Boundary markers: The instructions do not specify explicit delimiters or warnings to ignore instructions embedded within the source materials. 3. Capability inventory: The agent performs file write operations (.html) and may use browser tooling to inspect the results. 4. Sanitization: No explicit sanitization or validation of the content extracted from repository files is mentioned before it is synthesized into the generated code.

html