lugia19/Claude-Usage-Extension

Claude Usage Tracker

What it solves

Users of Claude.ai often want to monitor how much of their usage quota they have left. This extension provides a way to track token consumption across conversations, including from files, projects, chat history, system prompts, and MCPs/integrations (with some limitations). It helps users avoid hitting limits unexpectedly.

How it works

The extension calculates token usage from various sources: uploaded files, project knowledge, custom instructions, message history (including tool results), system prompts, and most MCP integrations. It calculates tokens either via Anthropic's API (if you provide your key) or using gpt-tokenizer. It also syncs usage data across devices using Firebase, after fetching your organization ID from claude.ai. The extension also shows UI elements with tooltips explaining details like length, cost, estimate, and caching status.

Who it's for

Anyone who uses Claude.ai and wants to keep track of their token usage, especially those who are concerned about hitting their quota. It's a browser extension available for Chrome and Firefox, and also has a desktop client installer for MacOS/Windows.

Highlights

  • Tracks token consumption from multiple sources: files, projects, personal preferences, message history, system prompts, and most MCPs/integrations.
  • Provides token calculation via Anthropic's API or gpt-tokenizer.
  • Syncs usage data across devices using Firebase (with organization ID from claude.ai).
  • UI includes tooltips for detailed info on length, cost, estimate, and caching status.

Related

  • Project
    lllucccian/Deekseep

    An LSPosed/Xposed module for the official DeepSeek Android app that provides a compatibility layer and additional settings to enhance the mobile user experience.

  • Project
    tiktoken-go/tokenizer

    A pure Go port of OpenAI's tokenizer for encoding and decoding text into tokens compatible with OpenAI's models.

  • Project
    tbckr/sgpt

    A Go-based command-line interface for interacting with OpenAI and other compatible LLMs to generate shell commands, code, and text responses directly in the terminal.

  • Project
    SlowLow999/UltraBr3aks

    UltraBr3aks is a repository that shares prompt-based jailbreak techniques for major AI models like GPT, Claude, and Gemini. It documents various attack methods, such as Attention-Breaking and token injection, to help researchers study and understand LLM safety vulnerabilities.