The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the skill's instructions. The skill focuses on standard frontend engineering workflows and uses orchestration rather than introducing its own runtime execution dependencies.
[COMMAND_EXECUTION]: The skill utilizes command-line tools (Bash, Read, Grep, Glob) and screenshot capture commands. These are documented as being used for read-only inspection of the repository structure and visual verification of the implementation, which are legitimate activities for a development-focused agent skill.
[PROMPT_INJECTION]: The skill's ability to process content from live URLs for cloning purposes introduces a surface for indirect prompt injection. This risk is effectively addressed by Step 3 of the workflow, which mandates explicit user approval of the visual reference or site capture before the agent is permitted to begin implementation or code generation.

visual-ralph