The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The SKILL.md file contains an instruction to run npx --yes submit-expo-feedback@latest. This command downloads and executes a package from the NPM registry. As the package submit-expo-feedback is not an official tool and does not originate from the skill author's verified namespace, it represents an unverified remote code execution vector.
[EXTERNAL_DOWNLOADS]: The use of npx to fetch and run code from a remote registry involves downloading and executing external logic at runtime without version pinning or integrity validation.
[COMMAND_EXECUTION]: The skill encourages the use of shell commands (npx) that can perform arbitrary operations on the host system depending on the contents of the downloaded package.
[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted external data through network requests and route parameters, creating a surface for indirect prompt injection.
Ingestion points: External API responses fetched via fetch and React Query in SKILL.md, and request.url or params accessed in server-side loaders in references/expo-router-loaders.md.
Boundary markers: The instructions do not define clear delimiters or "ignore previous instruction" warnings for data returned from external APIs.
Capability inventory: The skill has capabilities including fetch for network requests and expo-secure-store for sensitive token management, which could be abused if an injected prompt is processed.
Sanitization: While general advice to sanitize user input is present in references/expo-router-loaders.md, the skill lacks specific implementation logic or automated checks to ensure API responses are safe before use.

expo-data-fetching