Simplifying Control Flow

Simplifying Control Flow

Overview

Nested conditionals are hard to understand and error-prone. Flatten them.

Core principle: Nesting depth < 3 levels. Use early returns, table-driven methods, or extracted conditions.

Baseline Violation

Agents create nested if/else for multi-condition logic:

❌ Nested (baseline):

def calculate_discount(order_amount, is_vip):
    if is_vip:
        if order_amount > 1000:
            return 0.20
        elif order_amount > 500:
            return 0.15
    else:
        if order_amount > 1000:
            return 0.10
        elif order_amount > 500:
            return 0.05
    return 0.0

Problems: Duplicated logic, hard to see all tiers, adding tier requires finding nesting spot.

Three Techniques

1. Flatten with Combined Conditions

✅ All at same level:

def calculate_discount(order_amount, is_vip):
    if is_vip and order_amount > 1000: return 0.20
    if is_vip and order_amount > 500: return 0.15
    if order_amount > 1000: return 0.10
    if order_amount > 500: return 0.05
    return 0.0

2. Table-Driven (Best for Data)

✅ Business rules as data:

DISCOUNT_TIERS = [
    (1000, 0.20, 0.10),  # min_amount, vip_rate, regular_rate
    (500,  0.15, 0.05),
]

def calculate_discount(order_amount, is_vip):
    for min_amount, vip_rate, regular_rate in DISCOUNT_TIERS:
        if order_amount > min_amount:
            return vip_rate if is_vip else regular_rate
    return 0.0

When to use: Pricing tiers, status transitions, configuration-driven logic.

3. Extract Complex Conditions

✅ Named boolean for clarity:

def is_eligible(user, minimum):
    return (user.age >= 18 and user.verified_email and
            user.balance > minimum and not user.suspended)

if is_eligible(user, minimum_purchase):
    allow_purchase()

When to use: Complex boolean expressions, reused conditions.

Quick Reference

Guard Clauses

Baseline showed agents already use these well:

def validate(data):
    if not data:
        return False, "data required"  # Early return
    if data.amount <= 0:
        return False, "amount must be positive"  # Early return
    # Main logic here (no nesting)

Keep using this pattern for validation and error cases.

Red Flags

• Nesting depth > 3
• Can't see matching braces without scrolling
• Duplicate conditions in nested blocks
• Adding new case touches multiple nesting levels

Fix: Flatten or use table-driven.

Real-World Impact

From baseline:

• Agents created 2-level nesting for 4 discount tiers
• With table-driven: All tiers visible, easy to add/modify

Integration with Other Skills

For complex functions: See skills/keeping-routines-focused - extract when nesting gets deep

For reducing complexity: See skills/architecture/reducing-complexity - simpler control flow = less complexity