The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains purely technical instructions for mirroring design system patterns. No bypass attempts or safety-override patterns were found.
[DATA_EXFILTRATION]: No hardcoded credentials or unauthorized network operations are present. The file path references are standard for a software development environment.
[REMOTE_CODE_EXECUTION]: No remote script downloads or unsafe command executions from external sources were identified. Tooling references are local to the repository.
[OBFUSCATION]: The content is clear and readable with no signs of hidden instructions or encoding.
[INDIRECT_PROMPT_INJECTION]: While the skill instructs the agent to treat an external directory as a source of truth for patterns, this is a standard configuration task and does not involve executing unvalidated instructions from that source.

shadcn-parity