Skip to content

Quickstart

Get your first rule working in 5 minutes.

Prerequisites: - Python 3.10+ with pydantic>=2.8 - Installation complete

Your First Rule

Create a Python script:

from fluxrules import Rule, PhreakEngine

# Create a rule
rule = Rule(
    name="high_value_transaction",
    domain="fraud_detection",
    condition_dsl={
        "type": "condition",
        "field": "amount",
        "op": ">",
        "value": 5000,
    },
    action="manual_review",
    priority=100,
)

# Create an engine and load the rule
engine = PhreakEngine()
engine.load_rules([rule])

# Evaluate a fact
fact = {"amount": 6000}
result = engine.evaluate(fact)

print(f"Matched rules: {result.fired_rules}")
print(f"Actions: {result.actions}")

Output: 

Matched rules: [1]
Actions: ['manual_review']

Complex Conditions (AND/OR/NOT)

Rules can have nested boolean logic:

rule = Rule(
    name="risky_payment",
    domain="fraud_detection",
    condition_dsl={
        "type": "group",
        "op": "AND",  # All children must be true
        "children": [
            {"type": "condition", "field": "amount", "op": ">", "value": 2000},
            {
                "type": "group",
                "op": "OR",  # At least one child must be true
                "children": [
                    {"type": "condition", "field": "country", "op": "in", "value": ["NG", "SN"]},
                    {"type": "condition", "field": "ip_risk_score", "op": ">=", "value": 70},
                ],
            },
            {
                "type": "not",  # This condition must be false
                "condition": {"type": "condition", "field": "verified_user", "op": "==", "value": True},
            },
        ],
    },
    action="block_payment",
    priority=100,
)

engine = PhreakEngine()
engine.load_rules([rule])

# This fact matches all conditions
fact = {"amount": 3000, "country": "NG", "verified_user": False}
result = engine.evaluate(fact)
print(result.fired_rules)  # [1]

Multiple Rules

Load multiple rules at once:

rules = [
    Rule(
        name="high_value",
        condition_dsl={"type": "condition", "field": "amount", "op": ">", "value": 5000},
        action="manual_review",
        priority=100,
    ),
    Rule(
        name="high_risk_country",
        condition_dsl={"type": "condition", "field": "country", "op": "in", "value": ["NG", "SN"]},
        action="block_payment",
        priority=110,
    ),
]

engine = PhreakEngine()
engine.load_rules(rules)

fact = {"amount": 3000, "country": "NG"}
result = engine.evaluate(fact)

print(f"Matched: {result.fired_rules}")  # [2] (high_risk_country)
print(f"Actions: {result.actions}")       # ['block_payment']

Learn More

Next: Try the Full Learning Path

All examples use a shared payment-risk dataset:

cd examples
python 00_getting_started.py    # ~2 min
python 02_complex_conditions.py # ~2 min
python 10_pluggable_engines.py  # ~2 min

See examples.md for the complete 30-minute learning path.