The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to offer and potentially execute 'npx getdesign@latest add '. This command downloads and runs code from the NPM registry provided by a third party ('VoltAgent'). Since 'VoltAgent' is not a pre-verified trusted vendor, this pattern represents a remote code execution risk.\n- [COMMAND_EXECUTION]: The skill includes shell command strings ('npx getdesign...') for project setup and brand preset integration as part of its recommended workflow.\n- [EXTERNAL_DOWNLOADS]: The skill references and fetches design assets and code from an external repository ('github.com/VoltAgent/awesome-design-md') via a CLI tool, bypassing standard dependency verification.\n- [PROMPT_INJECTION]: The skill exhibits a significant surface for indirect prompt injection by ingesting and acting upon user-supplied external data.\n
Ingestion points: User-provided repository URLs and source code pastes in SKILL.md ('Source repo as reference' section), and user-supplied screenshots in references/mode-screenshot-iteration.md.\n
Boundary markers: There are no explicit instructions or markers provided to the agent to distinguish between legitimate design data and potentially malicious instructions embedded within the ingested code or visual evidence.\n
Capability inventory: The agent has the capability to execute shell commands ('npx') and generate/modify project files based on the ingested content.\n
Sanitization: The skill does not define any validation or sanitization procedures for the external code or screenshot data before it is processed by the agent.

ui