A specialized utility to summarize per-model usage and costs for Codex and Claude via the CodexBar CLI.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install model-usage-skill
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install model-usage-skill using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Model Usage for CodexBar?
The model-usage skill is a powerful diagnostic tool designed for developers who need granular insights into their AI model expenditures. By leveraging CodexBar's local cost logs, this skill enables users to extract and summarize cost data for specific providers like Codex and Claude. It is a vital component of the Openclaw Skills ecosystem, offering both high-level summaries and deep-dive breakdowns into which specific models are consuming the most budget. This allows for better financial oversight and optimization of AI-driven development workflows.
Model Usage for CodexBar Use Cases
- Monitoring daily spend across different AI models in real-time.
- Identifying which specific Claude or Codex model is responsible for the highest costs.
- Exporting model-level usage data to JSON for custom reporting or integration.
- Auditing historical cost logs provided by CodexBar to track usage trends.
How Model Usage for CodexBar Works
- The skill triggers the CodexBar CLI to fetch the latest cost data in JSON format for a specific provider.
- It executes a local Python script to process the modelBreakdowns and modelsUsed arrays within the local logs.
- The logic identifies the current model by looking at the most recent daily entries or allows for a full breakdown of all models historically.
- The processed data is then formatted into a readable text summary or a structured JSON response for the user to review.
Model Usage for CodexBar Setup
To use this among your Openclaw Skills, ensure you have CodexBar installed on macOS.
Install CodexBar via Homebrew:
brew install --cask steipete/tap/codexbar
The skill uses a Python script located in the scripts directory. Ensure your environment has Python installed to execute the following:
python scripts/model_usage.py --provider codex --mode current
Model Usage for CodexBar Data Schema & Taxonomy
The skill processes JSON data structures returned by the CodexBar CLI and organizes them for analysis.
| Field | Description |
|---|---|
| provider | Specifies the AI provider, such as codex or claude. |
| mode | Defines the scope of analysis, either current or all. |
| format | Determines output style, supporting text or json. |
| modelBreakdowns | An array containing daily cost records per model. |
| modelsUsed | A fallback array used when specific breakdowns are missing. |
Model Usage for CodexBar Advanced Features
- Automated fall-back logic that uses modelsUsed when specific modelBreakdowns are unavailable in the source JSON.
- Support for standard input (stdin) piping, allowing for complex CLI automation and scriptable workflows.
- Custom model overrides using the --model flag to target specific LLMs for cost extraction.
- Pretty-print JSON output for seamless integration with other Openclaw Skills and automated data pipelines.
SKILL.md
Loading
Related Openclaw Skills
Daily Briefing
An automated morning assistant that aggregates weather, calendar events, news, and AI insights into a concise iMessage delivery.
WeChat Daily Report Generator
An automated workflow that analyzes WeChat chat history to generate AI-powered, mobile-optimized daily report images.
Social Trends Analyst
A specialized tool for analyzing real-time social media hot spots and viral trends across major Chinese social platforms.
Safe Config Modifier
A standardized workflow tool for the safe, validated, and backed-up modification of OpenClaw configuration files.
Nate Jones Second Brain
A robust personal knowledge architecture using Supabase and OpenRouter to turn AI agents into long-term memory systems.
Fly.io (flyctl) CLI
A secure interface for managing Fly.io applications, databases, and deployments using the flyctl command-line tool.
Featured*
