The Agent Skills Directory

[SAFE]: The skill provides guidance on using createServerFn for server-side RPCs, including type-safe validation and response handling.- [SAFE]: Extensive security warnings are provided to prevent common developer errors, such as bypassing authentication by relying solely on route guards, leaking data through isomorphic loaders, or using incorrect cache headers for sensitive data.- [SAFE]: The documentation encourages the use of structured validation (e.g., Zod) for all external data ingestion points, such as FormData and API request headers.- [PROMPT_INJECTION]: The skill documents a surface for indirect prompt injection as it facilitates the ingestion of untrusted data through validator functions and FormData which are then processed by server-side database operations. Ingestion points: validator blocks and FormData processing in server function handlers (SKILL.md). Boundary markers: Explicit recommendations for Zod validation are present, though specific delimiters for LLM contexts are absent. Capability inventory: Database read/write operations (db.query, db.delete). Sanitization: The skill mandates schema validation using Zod to ensure data integrity.

server-functions