The Agent Skills Directory

[PROMPT_INJECTION]: The instructions are entirely focused on the professional role and technical responsibilities of a frontend specialist. No override markers, safety bypass attempts, or jailbreak patterns were identified.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or unauthorized network operations are present in the skill content.
[INDIRECT_PROMPT_INJECTION]: The agent possesses tools for file manipulation and shell access (Bash, Write, Read, MultiEdit), which are standard for development tasks. While these tools allow the agent to ingest external data (source code), the instructions do not contain unsafe interpolation or patterns that exploit this surface. As these capabilities are essential to the primary purpose of code generation and optimization, the risk is considered low and managed through standard agent guardrails.
[METADATA_POISONING]: Metadata fields such as author, version, and description are consistent with the skill's stated purpose as a ported tool from the Claude Code Marketplace. No deceptive or misleading instructions were found in the metadata sections.

frontend-developer