Model Usage & Cost

OpenClaw contributorsFinanceOpenClaw

Summarize local CodexBar cost logs by model for Codex or Claude.

View source

Illustrative Model Usage & Cost workflow: CodexBar cost JSON; example output includes Current model, All-model breakdown, Recorded cost totals.

Set up Model Usage & Cost

This skill ships with OpenClaw. Inspect its setup and check the required tools in the runtime where OpenClaw runs.

Inspect the bundled skill

openclaw skills info model-usage

Check runtime requirements

openclaw skills check

These commands inspect skill availability; they do not install dependencies or connect accounts. Follow the source’s setup requirements in your own environment.

What this skill helps with

OpenClaw’s Model Usage skill reads CodexBar’s local cost data and produces per-model summaries for Codex or Claude. Its bundled Python script supports a current-model view and a breakdown across all models.

When to use it

Use it to inspect the cost of local coding-agent usage, compare models in a recorded period or turn an exported cost log into a structured summary.

What to bring

Choose the provider and current or all-model mode. Supply local CodexBar logs or exported JSON; the source supports live reads on macOS and Linux and file-based input where Python is available.

What to expect

Expect text or JSON containing cost values by model. The source says the data does not split token counts by model. These local cost records are not a verified account invoice; the chart is illustrative.

How the skill works

Read the cost source

Use CodexBar’s provider-specific JSON export or a supplied JSON file. Keep the period and source of the data visible.

Choose the breakdown

Select the current-model rule, a named model or all-model mode. The source defines how it chooses the current model from the latest daily record.

Report what the log supports

Present recorded model costs in text or JSON. Do not infer a model-level token breakdown that the upstream data does not contain.

Related skills and guides

Model Usage & Cost skill questions

Can it show tokens per model?

The source says CodexBar’s output used here supplies per-model cost values without a per-model token split.

Can I use an exported file?

Yes. The Python summarizer accepts exported CodexBar JSON through its input option, including file or standard input.

Find the next skill for your workflow

Explore more task-specific skills, compare their requirements and review their original sources.

Browse Agent Skills