optimize-costs

/architect:optimize-costs

Analyzes current infrastructure costs and recommends 5-10 savings opportunities. For each saving, shows: monthly savings, trade-off, effort to implement, and ROI.

Instead of "cut costs anywhere", users get: "Here are 8 ways to save $3k/month. Ranked by savings. Pick which you're comfortable with."

Trigger

/architect:optimize-costs
/architect:optimize-costs [--target-monthly 3000]     # hit specific budget
/architect:optimize-costs [--target-percent -20]      # reduce costs by 20%
/architect:optimize-costs [--exclude database]        # don't touch certain services
/architect:optimize-costs [--show-all]                # show all 20+ possible optimizations

Purpose

Infrastructure costs creep upward:

• Database grows, per-query cost increases
• Traffic spikes, compute needs double
• Monitoring gets more verbose (logs cost more)
• You suddenly realize you're paying $15k/month

This command stops the creep. It finds low-hanging fruit ($500-5k/month savings) without hurting product.

Input

Context loading: Read architecture-output/_state.json.cost_estimate first if it exists — fall back to architecture-output/cost-estimate.md for detail not in _state.json.

If neither exists and the user has not supplied current spend interactively: "I need a cost baseline to optimize from. Run /architect:cost-estimate first, then come back here."

Required: Current cost estimate or actual spend

{
  "current_monthly_cost": 8500,
  "target_monthly_cost": 5500,      // optional: hit this budget
  "current_dau": 100000,            // expected users
  "scaling_plan_months": 6,         // when expect to double users?
  "can_sacrifice": {
    "latency": false,               // can we be slower?
    "uptime": false,                // can we have more downtime?
    "features": false,              // can we cut features?
    "scale": true                   // can we handle less traffic?
  }
}

Output

1. architecture-output/cost-optimization-plan.md