🖥️
Peter Steinbergerv1.0.0
Model Usage
Summarize per-model AI cost usage from CodexBar's local logs — supports Codex and Claude, current or all-model breakdowns.
Peter Steinbergerv1.0.0ProductivityAI PoweredAutomationCLIDeveloper Tool
Connecting to VM...
Connecting to VM...
npx clawhub@latest install model-usage69Stars
19.9kDownloads
9Current Installs
797All-time Installs
v1.0.0Version
Feb 26, 2026Updated
Platforms
darwin
Model Usage Skill Overview
Model Usage is a skill that reads CodexBar's local cost logs and produces per-model spending summaries for Codex or Claude. It can report on the single most-recently-used model ("current") or aggregate costs across every model in your history. Output is available as human-readable text or structured JSON, making it easy to pipe into dashboards or scripts.
How to Use It
Step 1: Run in your terminal or install this skill on MyClaw
npx clawhub@latest install model-usageor
Click the Install button at the top of this page for one-click setup
When to Use Model Usage
Best Fit
- You use CodexBar on macOS and want to know which AI model is costing the most right now.
- You need a scriptable, machine-readable breakdown of model-level costs from CodexBar's JSON output.
- You want to compare spending across all models (e.g.,
gpt-4ovsclaude-3-5-sonnet) over your usage history. - You want to feed model cost data into another tool or CI check via
--format json --pretty.
When Not to Use
- You are on Linux — CodexBar CLI support for Linux is not yet documented and this skill is macOS-only.
- You need token-level breakdowns per model — CodexBar's output provides cost only, not token splits.
- You are not using CodexBar at all — this skill has no value without CodexBar's local cost logs.
Key Features
Current-model detection
Automatically identifies the most recently used model by finding the latest daily row containing modelBreakdowns and selecting the highest-cost model. Falls back to modelsUsed when breakdowns are absent, and supports a manual --model <name> override.
Full model breakdown (all-models mode)
Aggregates cost data across every model in your CodexBar history with --mode all, giving you a complete picture of where your AI spend is going.
Codex and Claude provider support
Works with both --provider codex and --provider claude, covering the two main providers tracked by CodexBar.
Flexible input sources
Can fetch data live via codexbar cost --format json, read from a saved file with --input /path/to/cost.json, or accept piped stdin with --input -.
Text and JSON output formats
Returns human-readable text by default. Pass --format json --pretty to get structured JSON output suitable for scripting or further processing.
Use Cases
Daily cost check for active model
Run --provider codex --mode current each morning to instantly see which model you used most recently and what it cost, without digging through raw JSON logs.
Monthly model-spend audit
Use --mode all --format json --pretty to export a full model-by-model cost breakdown, then import it into a spreadsheet or cost-tracking dashboard.
Scripted cost alerting
Pipe the JSON output into a shell script or CI job that checks whether any model's cost exceeds a threshold, enabling automated budget guardrails.
Comparing Codex vs Claude spend
Run the skill twice — once with --provider codex and once with --provider claude — to compare total spending across both providers side by side.
Connecting to VM...
npx clawhub@latest install model-usage69Stars
19.9kDownloads
9Current Installs
797All-time Installs
v1.0.0Version
Feb 26, 2026Updated
Platforms
darwin
FAQ
Reviews
0 reviewsLog in to write a review
No reviews yet. Be the first to share your experience!