Introduction
Evalution is a local-first playground for editing and running your Vercel AI SDK prompts directly in your codebase.
Why in your codebase?
Your prompts are code. They reference your types, your helpers, and your real application context. Evalution runs them right where they live so what you evaluate is exactly what runs in production.
// simple.prompt.ts
import { prompts } from "@evalution/vercel-ai-sdk";
export default prompts(
{ id: "Simple" },
({ openai }) => ({
simplePrompt: () => ({
model: openai("gpt-5.4-mini"),
system: "You are a helpful assistant",
messages: [{ role: "user", content: "Hello!" }],
}),
}));