The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill promotes the use of npx to download and run the typeui.sh CLI tool. This pattern involves executing code from an unverified external source directly on the host machine.
[PRIVILEGE_ESCALATION]: The troubleshooting documentation explicitly suggests running the CLI with sudo or using sudo chown to bypass permission errors. Granting elevated privileges to an untrusted third-party tool is a high-risk security practice.
[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to pull remote SKILL.md files from a third-party API and provide them as instructions to AI agents (Claude, Cursor, Codex). This introduces a supply chain risk where malicious content in the registry could inject unauthorized instructions.
Ingestion points: design-skills/*/SKILL.md files downloaded from typeui.sh/api/skills/.
Boundary markers: Absent; no delimiters or warnings are used to prevent the agent from obeying embedded instructions.
Capability inventory: The agent uses these files for code generation and project configuration (e.g., Cursor rules).
Sanitization: Absent; content is ingested directly from the remote API.
[EXTERNAL_DOWNLOADS]: The skill instructs users to install and run the typeui.sh package from a non-trusted repository.
[COMMAND_EXECUTION]: The typeui.sh tool performs automated file system modifications and network requests to download and manage external design assets.

awesome-design-skills-registry