The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx ui-skills get <skill-slug> to retrieve external content based on hardcoded third-party identifiers (e.g., jakubkrehel/better-layout, ibelick/fixing-accessibility). These external sources are not part of any established trusted vendor list and could potentially deliver malicious content that the agent then processes or executes.
[EXTERNAL_DOWNLOADS]: The protocol involves fetching external design "lenses" from remote repositories using the npx utility, introducing dependencies on external, unverified third-party content.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands, specifically npx for data retrieval and open for launching the generated HTML mockups in the user's browser.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface:
Ingestion points: Reads meta/DESIGN.md and external content via npx ui-skills.
Boundary markers: The instructions do not specify the use of delimiters or "ignore embedded instructions" warnings for the ingested data.
Capability inventory: Includes file-write capabilities, network operations via npx, and command execution via open.
Sanitization: There is no mention of sanitization or escaping of external content before it is interpolated into the generated HTML mockups.
[DYNAMIC_EXECUTION]: The skill dynamically generates one-file HTML artifacts from a mix of project data, external design principles, and user input. The subsequent use of the open command to view these files in a browser context means any injected scripts or malicious HTML attributes would be executed in the user's local environment.

generate-mockups