Claude Usage Tracker is a local-first tool designed to automatically discover and aggregate your Claude AI usage from over 9 different local development tools. It scans known data directories, parses JSONL/log files, and calculates costs based on Anthropic's model-specific pricing. The tool then presents this information in an interactive, dark-themed dashboard powered by Chart.js.
This tool is ideal for developers, engineers, and anyone using Claude AI through various local integrations like OpenClaw, Claude Code CLI, Cursor, or Aider. It helps users understand their spending patterns, identify peak usage times, and break down costs by different Claude models (Opus, Sonnet, Haiku). A key benefit is its privacy-focused approach: all data processing and visualization happen locally on your machine, with no cloud uploads or telemetry.
The Claude Usage Tracker provides a comprehensive overview of AI spending, offering daily, weekly, monthly, and all-time cost tracking, per-model analytics, and heatmaps to visualize usage patterns. It also includes a detailed session log, multi-criteria filtering, and monthly cost projections, enabling users to manage and optimize their Claude AI expenditures effectively.
Which specific local development tools does Claude Usage Tracker support for aggregating AI usage?
Claude Usage Tracker automatically detects and aggregates usage from OpenClaw/Clawdbot, Claude Code CLI, Claude Desktop, Cursor, Windsurf, Cline, Roo Code, Aider, and Continue.dev.
How does the tracker calculate the cost for Claude AI usage, and which models are supported?
Costs are calculated using Anthropic's per-million-token pricing for input, output, cache write, and cache read. It supports all current and upcoming model families including Opus (4.5/4.6, 4.0/4.1), Sonnet (3.5/3.7/4.0/4.5/4.6), and Haiku (4.0/4.5, 3.0/3.5).
Can I use Claude Usage Tracker on operating systems other than macOS?
While a standalone macOS app is provided, users on other operating systems can run the tool in 'Browser Mode' by executing node collect-usage.js and then python3 -m http.server 8765 to view the dashboard in a web browser.
What kind of data visualization and filtering options are available in the dashboard?
The dashboard offers daily, weekly, monthly, and all-time cost tracking, per-model cost breakdowns, usage heatmaps (Peak Hours and Peak Days), and an expandable session log. Users can filter data by source tool, Claude model, date range, and minimum cost.