tokenmeter

What This Skill Does

The tokenmeter skill for OpenClaw is a powerful, privacy-focused utility designed to bring transparency to your AI usage. As LLM interactions become a staple of daily workflows, tracking token consumption and associated costs is essential for both individual power users and enterprise environments. This tool monitors your AI interactions across all OpenClaw instances, logging data locally into a secure SQLite database (~/.tokenmeter/usage.db). Because it operates entirely locally without telemetry or cloud-syncing, your usage patterns remain strictly private. Beyond simple monitoring, tokenmeter acts as a financial intelligence tool, helping you calculate real-time cost estimates, identify which models drive your expenses, and audit your usage patterns to catch potential overages or inefficiencies before they impact your budget.

Installation

The OpenClaw bot manages the entire installation lifecycle for this skill automatically. When you first invoke a tokenmeter command, the bot checks for the existence of the utility in your ~/clawd/ directory. If it is missing, the bot performs a automated setup: it clones the repository from https://github.com/jugaad-lab/tokenmeter.git, initializes a dedicated Python virtual environment to manage dependencies, and installs the necessary packages. Once the initial setup concludes, the bot simply activates the environment for subsequent requests. Users do not need to perform any manual configuration, path management, or dependency resolution, making it a true 'install-and-forget' experience.

Use Cases

• Expense Auditing: Periodically review your spending to ensure you remain within budget, especially when utilizing high-end models like Claude-Opus.
• Subscription Value Tracking: Compare your current API-based usage against the potential savings of a fixed-price 'Max' subscription, proving ROI on subscription costs.
• Optimization Analysis: Identify 'model bloat' by detecting if you are using expensive models for simple tasks that could be handled by faster, cheaper alternatives.
• Unified Reporting: Aggregate data from multiple OpenClaw instances into a single source of truth for your total AI footprint.

Example Prompts

1. "tokenmeter, how much did I spend on AI models during the last week?"
2. "Can you show me a breakdown of my costs by model to see where my tokens are going?"
3. "Run a report to compare my current API usage against what I would pay under the Max plan savings."

Tips & Limitations

To get the most out of tokenmeter, run the import command regularly to ensure your data stays fresh. While the tool is highly accurate for tracked sessions, it is limited to the data accessible within your OpenClaw session files. It does not track usage from third-party tools outside the OpenClaw environment. For the best security, the local SQLite file should be included in your standard system backup routine if you wish to retain long-term historical records.