The Agent Skills Directory

[SAFE]: The skill is a set of technical instructions and code snippets for frontend development (React/Next.js). It does not contain executable scripts, hidden commands, or external network requests.
[INDIRECT_PROMPT_INJECTION]: The skill describes patterns for ingesting data from URL query parameters (searchParams.get()), which is untrusted external input.
Ingestion points: URL parameters accessed via useSearchParams or the native URLSearchParams API in SKILL.md.
Boundary markers: None required for this UI-state context, though the instructions recommend keeping the URL as the source of truth.
Capability inventory: The skill only performs UI state updates and browser history navigation. No file writes, network operations, or shell executions are present.
Sanitization: The skill explicitly includes a 'Validation and Type Safety' section (Section 8) that demonstrates how to validate, type-check, and provide fallbacks for URL parameters before they are used, effectively mitigating risks from malformed or malicious inputs.

deep-linking