claudeusage-mcp
Provides real-time visibility into Claude Pro and Max subscription usage limits directly within Claude Code by utilizing local OAuth tokens. It enables users to monitor session and weekly usage across different models and receive alerts regarding rate-limiting status.
README
claudeusage-mcp
An MCP server that gives you real-time visibility into your Claude Pro/Max subscription usage — directly inside Claude Code.
No API keys. No scraping. No browser automation. It reads the OAuth token that Claude Code already stores on your machine and calls Anthropic's usage endpoint to get the exact same data shown on claude.ai/settings/usage.
What it shows
| Metric | Description |
|---|---|
| Current Session | 5-hour burst window utilization with reset countdown |
| Weekly Usage - All Models | 7-day rolling limit across all models |
| Weekly Usage - Opus Only | 7-day Opus-specific usage (Max plans) |
| Weekly Usage - Sonnet Only | 7-day Sonnet-specific usage |
| Extra Usage | Enabled/disabled, monthly limit, credits used |
MCP tool output (get_usage)
=== Claude Plan Usage ===
Current Session (5-hour window)
████████░░░░░░░░░░░░ 37%
Resets in 2h 14m (Sat Mar 22 07:59PM)
Weekly Usage - All Models (7-day) [HIGH]
██████████████████░░ 89%
Resets in 4d 3h (Wed Mar 26 05:00PM)
Weekly Usage - Opus Only
N/A (not on your plan)
Weekly Usage - Sonnet Only
░░░░░░░░░░░░░░░░░░░░ 1%
Resets in 4d 3h (Wed Mar 26 08:59PM)
Extra Usage: Disabled
--- Color Grading ---
0-49% = GREEN (safe) | 50-79% = ORANGE (moderate) | 80-99% = RED (high) | 100% = RATE LIMITED
(live data)
Rate-limited state
When any metric hits 100%, the output changes to show reset date and time:
Weekly Usage - All Models (7-day) [RATE LIMITED]
XXXXXXXXXXXXXXXXXXXX 100%
Resets in 1d (Mon Mar 24 05:00PM)
Statusline output
The included statusline script adds a third line to your Claude Code status bar with progress bars and color coding:
Usage: Session █░░░░░░░░░ 13% | Weekly █████████░ 94% | Sonnet █░░░░░░░░░ 8%
When rate-limited:
Usage: Session: RATE LIMITED, Resets in: 3h 37m (Sun Mar 22 07:59PM) | Weekly: RATE LIMITED, Resets in: 1d (Mon Mar 23 05:00PM)
Color grading
| Range | Color | Status |
|---|---|---|
| 0-49% | Green | Safe |
| 50-79% | Orange | Moderate |
| 80-99% | Red | High |
| 100% | Red + text | RATE LIMITED with reset datetime |
Tools
| Tool | Purpose |
|---|---|
get_usage |
Full dashboard — session, weekly, per-model, extra usage with progress bars |
get_session_usage |
Current 5-hour session window only |
get_weekly_limits |
7-day limits with per-model breakdown (All, Opus, Sonnet) |
check_rate_status |
Am I about to be rate-limited? Returns LOW / MODERATE / HIGH / RATE LIMITED |
Requirements
- Node.js >= 18
- Claude Code logged in with a Pro or Max subscription
- macOS or Linux (reads credentials from Keychain or
~/.claude/.credentials.json)
Installation
Option 1: claude mcp add (recommended)
git clone https://github.com/OrelliusAI/claudeusage-mcp.git
cd claudeusage-mcp
npm install && npm run build
claude mcp add claudeusage -- node $(pwd)/dist/index.js
Restart Claude Code. Then ask "what's my usage?" or type /usage.
Option 2: Manual config
git clone https://github.com/OrelliusAI/claudeusage-mcp.git
cd claudeusage-mcp
npm install && npm run build
Add to your MCP config (.claude.json or project .mcp.json):
{
"mcpServers": {
"claudeusage": {
"command": "node",
"args": ["/absolute/path/to/claudeusage-mcp/dist/index.js"]
}
}
}
Optional: Statusline integration
Add usage data to your Claude Code statusline (always visible at the bottom):
# If you DON'T have a statusline yet:
cp statusline.sh ~/.claude/usage-statusline.sh
chmod +x ~/.claude/usage-statusline.sh
Then set in ~/.claude/settings.json:
{
"statusLine": {
"type": "command",
"command": "bash ~/.claude/usage-statusline.sh"
}
}
If you already have a statusline, use statusline-combined.sh instead — it pipes your existing statusline and appends the usage line.
How it works
Claude Code (your session)
|
|-- Reads OAuth token from:
| |-- macOS Keychain ("Claude Code-credentials")
| +-- ~/.claude/.credentials.json (Linux/WSL)
|
|-- Calls: GET https://api.anthropic.com/api/oauth/usage
| Headers: Authorization: Bearer <your-oauth-token>
| anthropic-beta: oauth-2025-04-20
|
|-- Caches response for 60 seconds
| +-- On 429 (rate limit): serves stale cache
|
+-- Returns formatted usage data via MCP tools
Zero configuration. The OAuth token is created automatically when you sign into Claude Code with your Pro/Max subscription. No API keys, no environment variables, no .env files.
Authentication
This server uses the same OAuth token that Claude Code creates when you log in:
| Platform | Location |
|---|---|
| macOS | Keychain (service: Claude Code-credentials) |
| Linux / WSL | ~/.claude/.credentials.json |
Required OAuth scopes: user:inference, user:profile
Token issues?
If the server reports authentication errors:
- Your token may be expired — restart Claude Code to trigger a fresh OAuth login
- Token created via
claude setup-tokenor/loginmay have wrong scopes — delete credentials and restart Claude Code
Do NOT use API keys (sk-ant-api...). This server is for Pro/Max subscribers, not API users.
Rate limiting
Anthropic's /api/oauth/usage endpoint is aggressively rate-limited. This server handles it:
- Caching — responses cached for 60 seconds (MCP) or 120 seconds (statusline)
- Stale fallback — on 429, the last successful response is served
- Graceful errors — clear messages when the API is unavailable
Who is this for?
Claude Pro and Max subscribers who use Claude Code and want to see their plan usage without opening a browser. This is not for API users — if you have an API key (sk-ant-api...), use the Usage and Cost API instead.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.