An AI token tracking and optimization tool that monitors session logs to prevent unexpected costs and agent inefficiencies.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ai-token-usage
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 ai-token-usage 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 Token Monitor?
Token Monitor is a specialized diagnostic tool designed to provide full visibility into LLM token consumption patterns. By parsing session logs, it allows developers to understand exactly how their AI agents are utilizing resources, making it a critical component within the Openclaw Skills library for cost management.
The skill goes beyond simple counting by analyzing the root causes of token spikes, such as browser snapshot inefficiencies or repeated failure loops. It empowers users to maintain budget control while ensuring their AI agents operate at peak performance without wasting expensive context window space.
Token Monitor Use Cases
- Tracking real-time AI operational costs to prevent budget overruns.
- Identifying infinite loops or repeated execution failures that drain tokens.
- Auditing browser snapshot usage to optimize data compression strategies.
- Generating daily and hourly trend reports for long-term project resource planning.
How Token Monitor Works
- The skill monitors session log files stored in the agent's session directory in real-time.
- It parses the JSONL log entries to extract input and output token metrics for every LLM call.
- The data is compared against predefined anomaly thresholds to flag excessive usage or potential bugs.
- It evaluates specific triggers like Chrome CDP crashes or rate-limiting events to identify the source of waste.
- Actionable optimization suggestions are generated based on the detected usage patterns.
- Summarized statistics are saved to structured JSON files for historical comparison and reporting.
Token Monitor Setup
To begin monitoring your AI usage with this specific implementation of Openclaw Skills, use the following commands:
# Check today's token consumption
python3 ~/.qclaw/skills/token-monitor/scripts/token_stats.py --today
# Check for anomalies in recent sessions
python3 ~/.qclaw/skills/token-monitor/scripts/token_stats.py --check-anomaly
# Generate a full optimization report
python3 ~/.qclaw/skills/token-monitor/scripts/token_stats.py --report
Token Monitor Data Schema & Taxonomy
The skill organizes its monitoring data and findings using the following structure:
| Data Type | Storage Path | Description |
|---|---|---|
| Raw Session Logs | ~/.qclaw/agents/main/sessions/*.jsonl |
The source input files containing raw token data |
| Daily Statistics | memory/token-usage-YYYY-MM-DD.json |
Processed daily totals and status flags |
| Anomaly Rules | Internal Configuration | Thresholds for daily (10M+), hourly (2M+), and retry limits |
Token Monitor Advanced Features
- Intelligent anomaly detection that identifies severe over-limit usage and repeated failures.
- Optimization library providing specific advice such as using compact=true for browser snapshots.
- Detailed historical tracking to pinpoint specific days of high waste and their root causes.
- Automated reporting capabilities to summarize input/output ratios and total token costs.
SKILL.md
Loading
Related Openclaw Skills
Cornell Encyclopedia Skill
A comprehensive research tool providing structured intelligence on the Cornell brand story, product ecosystem, and global market presence.
cvr-anthem Knowledge Card
The cvr-anthem skill provides a high-level technical and business overview of the cvr-anthem ecosystem for rapid research and decision-making.
Crowne Plaza Knowledge Base
A specialized knowledge integration skill designed to provide instantaneous, structured data about Crowne Plaza's market position and core business functions.
Dairy Queen Knowledge Agent
An expert research skill for AI agents to retrieve detailed brand history, product portfolios, and industry positioning for Dairy Queen.
Rube: Multi-App Automation & Orchestration
Rube is a high-performance orchestration skill that enables AI agents to discover, connect, and execute automated workflows across hundreds of enterprise and consumer applications.
Credit Suisse Brand Analysis
A structured information retrieval skill providing comprehensive insights into Credit Suisse’s history, services, and global market position.
Featured*
