CriterionDecision Engine

Universal, deterministic, and explainable decisions for business-critical systems

Get Started

Try it Now

GitHub

🎯

Pure & Deterministic

Same input always produces the same output. No hidden state, no side effects, no surprises.

📋

Fully Explainable

Every decision comes with a complete audit trail. Know exactly which rule matched and why.

🔒

Contract-First

Zod schemas validate inputs, outputs, and profiles. Catch errors before they happen.

⚡

Profile-Driven

Parameterize decisions by region, tier, or environment without changing logic.

🧪

Testable by Design

Pure functions are easy to test. No mocks, no stubs, just inputs and outputs.

📦

Zero Dependencies

Only requires Zod as a peer dependency. Lightweight and focused.

Quick Example

typescript

import { Engine, defineDecision } from "@criterionx/core";
import { z } from "zod";

const riskDecision = defineDecision({
  id: "transaction-risk",
  version: "1.0.0",
  inputSchema: z.object({ amount: z.number() }),
  outputSchema: z.object({ risk: z.enum(["HIGH", "MEDIUM", "LOW"]) }),
  profileSchema: z.object({ threshold: z.number() }),
  rules: [
    {
      id: "high-risk",
      when: (input, profile) => input.amount > profile.threshold,
      emit: () => ({ risk: "HIGH" }),
      explain: (input, profile) =>
        `Amount ${input.amount} exceeds threshold ${profile.threshold}`,
    },
    {
      id: "low-risk",
      when: () => true,
      emit: () => ({ risk: "LOW" }),
      explain: () => "Amount within acceptable range",
    },
  ],
});

const engine = new Engine();
const result = engine.run(
  riskDecision,
  { amount: 15000 },
  { profile: { threshold: 10000 } }
);

console.log(result.data);  // { risk: "HIGH" }
console.log(result.meta.explanation);  // "Amount 15000 exceeds threshold 10000"