The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation suggests the installation of several well-known and trusted third-party libraries from the npm registry to support specific integrations. These include:
async-validator (for useAsyncValidator)
axios (for useAxios)
change-case (for useChangeCase)
universal-cookie (for useCookies)
drauu (for useDrauu)
focus-trap (for useFocusTrap)
fuse.js (for useFuse)
idb-keyval (for useIDBKeyval)
jwt-decode (for useJwt)
nprogress (for useNProgress)
qrcode (for useQRCode)
[INDIRECT_PROMPT_INJECTION]: The skill provides documentation for building features that ingest external or user-provided data. This represents a potential indirect prompt injection surface if the resulting applications process this data without adequate sanitization.
Ingestion points: Data enters the application context through composables like useFetch, useStorage, useClipboard, useUrlSearchParams, and useRouteQuery.
Boundary markers: The documentation uses standard JavaScript interpolation in examples (e.g., in references/computedAsync.md and references/useFetch.md) but does not explicitly enforce boundary markers for external content as it is a utility library guide.
Capability inventory: The skill guides the use of browser capabilities such as network requests (useFetch), persistent storage (useStorage), and clipboard access (useClipboard).
Sanitization: Sanitization is the responsibility of the application developer; the composables provide raw access to the underlying browser APIs.

vueuse-functions