The Agent Skills Directory

[SAFE]: The skill provides high-quality technical documentation and instructional boundaries for the agent. It strictly defines ownership (e.g., 'framework-owner-controlled-do-not-edit') and scope, ensuring the agent respects the boundaries between the build tool (Vite) and UI frameworks (React).
[DYNAMIC_EXECUTION]: The evaluation suite includes validator scripts (e.g., evals/validators/ssr-production-artifacts.mjs) that use dynamic import() to execute and verify the agent's output. This is a standard and appropriate testing technique for a developer-oriented skill to ensure generated configurations and adapters are functionally correct.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and migrate project files, which constitutes a potential attack surface. However, the instructions include detailed verification steps, deployment topology requirements, and explicit warnings against ignoring safety or semantic rules, which appropriately mitigates the risks associated with processing untrusted project data.

vite-guide