The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines templates for features that process external data from URL search parameters and form submissions, creating a theoretical surface for indirect prompt injection if the agent processes malicious input from these fields.
Ingestion points: Search parameters are managed via searchParamsCache (server-side) and useQueryStates (client-side) as seen in src/app/dashboard/<name>/page.tsx and src/features/<name>/components/orders-table/index.tsx. Form data is handled in feature components using TanStack Form.
Boundary markers: The provided instructions do not specify explicit delimiters for external data, but the architectural pattern relies on structured data management.
Capability inventory: The skill empowers the agent to generate and modify application code (file-write) and includes mock service layers that simulate network interactions.
Sanitization: The templates enforce strict validation using Zod schemas for forms and typed parsers for search parameters (nuqs), significantly mitigating risks from malformed input.
[EXTERNAL_DOWNLOADS]: The skill references standard, well-known development dependencies including @faker-js/faker, match-sorter, @tanstack/react-query, zod, nuqs, and recharts. These packages are standard in the React ecosystem and their usage here aligns with the skill's primary purpose.
[SAFE]: The skill instructions and associated reference guides (Charts, Forms, Mock API, Query Abstractions, Theming) align with modern developer workflows and Next.js 16 conventions. No malicious patterns such as obfuscation, credential harvesting, persistence mechanisms, or unauthorized privilege escalation were detected.

kiranism-shadcn-dashboard