Card Optimizer

Maximize credit card rewards by always using the right card for every purchase.

Data Location

• Skill logic: skills/card-optimizer/ (this file)
• User data: data/card-optimizer/
  • cards.json — card definitions, reward rates, spending estimates, category map

Card Database Schema

Each card in cards.json follows this structure:

{
  "id": "unique_id",
  "name": "Card Name",
  "issuer": "Issuer Name",
  "network": "visa|mastercard|amex|discover",
  "annual_fee": 95,
  "reward_type": "cashback|points|miles",
  "point_valuation_cpp": null,
  "transfer_partners": [],
  "notes": "Optional notes",
  "signup_bonus": {
    "amount": 200,
    "type": "cashback",
    "spend_requirement": 3000,
    "timeframe_months": 3,
    "earned": false
  },
  "categories": [
    {
      "category": "groceries",
      "rate": 6.0,
      "cap_amount": 6000,
      "cap_period": "yearly",
      "rate_after_cap": 1.0
    },
    {
      "category": "rotating",
      "rate": 5.0,
      "cap_amount": 1500,
      "cap_period": "quarterly",
      "rate_after_cap": 1.0,
      "quarterly_categories": {
        "Q1": ["gas", "ev_charging"],
        "Q2": ["groceries", "home_improvement"],
        "Q3": ["restaurants", "paypal"],
        "Q4": ["amazon", "target", "walmart"]
      },
      "activation_required": true
    },
    {
      "category": "everything_else",
      "rate": 1.0
    }
  ]
}

Point Valuations

For points/miles cards, store point_valuation_cpp (cents per point):

• Chase Ultimate Rewards: 1.0 cpp base, 1.25 cpp with Sapphire Preferred, 1.5 cpp with Sapphire Reserve
• Amex Membership Rewards: 1.0 cpp base, varies by transfer partner
• When comparing cards, multiply rate × point_valuation_cpp to get effective cashback equivalent

Category Map

The category_map in cards.json maps each spending category to the best card ID. This is the precomputed optimal assignment — recalculate when cards are added or removed.

Spending Estimates

To power ROI calculations, gap analysis, and new card recommendations, the user can optionally set estimated monthly spending per category in cards.json:

{
  "estimated_monthly_spending": {
    "groceries": 600,
    "gas": 200,
    "restaurants": 300,
    "amazon": 150,
    "streaming": 50,
    "everything_else": 500
  }
}

If no estimates are provided, the skill can still recommend cards per-purchase — it just can't run ROI or gap analysis. Ask the user to estimate during first-time setup.

Note: This skill does NOT track individual purchases. If the user wants detailed spending data, they should connect their bank accounts through a budgeting tool. These estimates are rough numbers for optimization calculations.

Purchase Optimizer

How to Recommend a Card

When the user asks "which card for [category]?" or "I'm buying [item]":