Model Usage Skill for OpenClaw

Track AI model costs from CodexBar CLI — per-model breakdowns for Codex or Claude.

Features

• 📊 Per-model cost tracking — See exactly which models cost you the most
• 💰 30-day summaries — Track spend over time
• 🔍 Current vs All — Check latest session or full history
• 📈 JSON/Text output — Scriptable or human-readable

Quick Start

# Current model (most recent day)
python scripts/model_usage.py --provider codex --mode current

# All models (full breakdown)
python scripts/model_usage.py --provider claude --mode all

# JSON output for scripting
python scripts/model_usage.py --provider claude --mode all --format json --pretty

Requirements

• CodexBar CLI installed (brew install steipete/tap/codexbar)
• Python 3.8+

Configuration

No config needed — reads from CodexBar's local cost logs automatically.

Output Example

Provider: codex
Models:
- gpt-5.2: $13.08

Provider: claude
Models:
- claude-opus-4-5: $53.47
- claude-sonnet-4-5: $2.81
- claude-haiku-4-5: $0.20

Use Cases

• Track which AI models are burning through your budget
• Generate usage reports
• Identify cost spikes
• Optimize model selection

License

MIT