Introduction

The Policy system is a rule engine that evaluates human-readable business rules against structured data. Instead of writing validation logic in code, you write policies in a natural-language DSL.

The Three Tiers

1. Engine

The engine is the core evaluator. It accepts a rule (policy text) and data (JSON), evaluates the rule against the data, and returns a result.

POST https://api.policy.keloran.dev/run
{
  "rule": "A **Person** gets approved if __age__ of **Person** is greater than 18.",
  "data": { "Person": { "age": 25 } }
}

2. Orchestrator

The orchestrator manages named policies. You create, version, and organize policies into groups. When you want to evaluate data, you reference a policy by name instead of sending the full rule text.

3. Dashboard

A web UI for creating and testing policies visually. It connects to the orchestrator and engine under the hood.

Why Use Policy DSL?

• Business rules change frequently — update a policy without redeploying code
• Stakeholder transparency — non-developers can read and verify the rules
• Audit trail — policies are versioned and traceable
• Separation of concerns — logic lives outside your application code
• Testable — every rule can be tested independently with sample data

How Rules Work

A policy rule has this structure:

A **Selector** outcome
  if condition
  [and/or condition]...
  .

• Selector — the JSON object being evaluated (e.g., **Person**)
• Outcome — what happens if all conditions pass (e.g., gets approved)
• Conditions — checks against properties of the data
• Period — every rule ends with .

The engine walks each condition, evaluates it against the data, and returns true (pass) or false (fail).