OpenClaw Cost Optimizer Skill

Overview

A community-created skill addressing one of the most common pain points in multi-provider AI agent setups: cost visibility and optimization.

The Problem

"The BIGGEST problem is no one knows what APIs are being called and how much it costs."

Solution Features

Token Usage Tracking

Real-time token consumption monitoring
Per-provider cost breakdown
Per-skill usage attribution
Historical trend analysis

Cost Optimization

Automatic model routing based on cost/performance
Alerts on unusual spending patterns
Budget thresholds and limits
Recommendations for cost reduction

Provider Comparison

Provider	Model	Cost per 1M tokens	Use Case
Anthropic	Haiku	$0.25	Routine tasks
Anthropic	Sonnet	$3.00	General tasks
Anthropic	Opus	$15.00	Complex reasoning
OpenAI	GPT-4o	$5.00	General tasks
Google	Gemini Flash	$0.35	Fast responses

Implementation

Configuration

{
  "cost_optimizer": {
    "enabled": true,
    "daily_budget": 10.00,
    "alert_threshold": 0.80,
    "track_by_skill": true,
    "optimize_routing": true
  }
}

Commands

/cost status          - Show current usage
/cost breakdown       - Per-skill breakdown
/cost optimize        - Get optimization suggestions
/cost alert set 5.00  - Set budget alert
/cost report daily    - Daily usage report

Optimization Strategies

1. Model Routing

Route routine tasks to cheaper models (Haiku, Gemini Flash)
Reserve expensive models (Opus) for complex tasks
Automatic fallback if cheaper model fails

2. Token Efficiency

Compact prompts
Response caching
Context window management

3. Batch Processing

Group similar requests
Process during off-peak hours
Async execution where possible

Real-World Impact

Before Optimization

Average: 20-40k tokens per request
Cost: ~$0.50-1.00 per complex task

After Optimization

Average: ~1.5k tokens per request
Cost: ~$0.05-0.10 per task
Savings: 80-90%

Installation

openclaw skills install cost-optimizer
openclaw secrets configure cost-optimizer
openclaw cost-optimizer init

References

Facebook Community (March 2026)
OpenClaw Cost Optimization Guide