The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill uses a TypeScript utility script executed via npx ts-node. No arbitrary command execution, shell spawning, or unsafe input evaluation (like eval or exec) was detected.
[DATA_EXFILTRATION] (SAFE): The script reads from a local JSON catalog file at a hardcoded relative path. It does not perform any network requests or have the capability to send data to external servers.
[PROMPT_INJECTION] (SAFE): The instructions in SKILL.md define triggers for design-related tasks and do not contain any instructions that attempt to bypass AI constraints or override system prompts.
[REMOTE_CODE_EXECUTION] (SAFE): The skill relies on local script execution. No patterns for downloading and executing untrusted remote code were found.
[INDIRECT_PROMPT_INJECTION] (INFO): While the skill processes untrusted user descriptions, its capabilities are limited to returning a JSON object with scores and keywords. It has no side-effect capabilities (e.g., file writing, API calls) that could be exploited via malicious input strings.

design-system-generator