The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends executing 'npx --yes submit-expo-feedback@latest' for reporting feedback. This package name does not follow the vendor's documented naming conventions (such as '@expo/' or 'expo-') and originates from an unverified source outside of the globally trusted organizations list. The use of the '@latest' tag without version pinning increases the risk of supply chain attacks.
[REMOTE_CODE_EXECUTION]: The suggested 'npx --yes' command facilitates the immediate download and execution of remote code. Since the package identity is not verified against the author's established resource patterns, it poses a risk of executing untrusted or malicious scripts on the user's system.
[PROMPT_INJECTION]: The skill documents a method for exposing native actions (like 'showAlert' or 'saveData') to a webview, which establishes a surface for indirect prompt injection. 1. Ingestion points: 'components/dom-component.tsx' receives data from the webview context via props. 2. Boundary markers: No delimiters or warnings to ignore embedded instructions are present in the examples. 3. Capability inventory: Native actions like 'Alert.alert' and data saving are exposed to the untrusted web environment. 4. Sanitization: No evidence of input validation or sanitization is provided, allowing potentially malicious web content to influence native application logic.

expo-dom