Skip to main content

​
Overview

CodexBar supports tracking usage limits across 15+ AI providers simultaneously. Each provider can be toggled independently in Settings → Providers, giving you complete control over which services appear in your menu bar.

​
Supported Providers

CodexBar integrates with the following AI services:

Codex

Local CLI RPC + optional web dashboard extras

Claude

OAuth API, browser cookies, or CLI fallback

Cursor

Browser session cookies for plan + usage

Gemini

OAuth-backed quota API via CLI credentials

Copilot

GitHub device flow + internal usage API

Antigravity

Local language server probe (experimental)

Droid (Factory)

Browser cookies + WorkOS token flows

z.ai

API token for quota + MCP windows

Kimi

Auth token for weekly quota + rate limit

Kiro

CLI-based usage via kiro-cli command

Vertex AI

Google Cloud gcloud OAuth with cost tracking

JetBrains AI

Local XML-based quota from IDE config

Augment

Browser cookie auth with auto keepalive

Amp

Browser cookie-based usage tracking

OpenRouter

API token for multi-provider credit tracking

​
Authentication Strategies

CodexBar uses multiple authentication methods depending on the provider:

​
Local CLI/RPC

  • Codex: JSON-RPC via codex command with PTY fallback
  • Claude: PTY scrape from claude CLI
  • Kiro: Usage command via kiro-cli
  • Gemini: OAuth credentials from Gemini CLI

​
Browser Cookies

  • Cursor, Droid, Amp: Reuses browser session cookies
  • Cookie source picker: Choose Automatic or Manual in Settings
  • Cached in Keychain: Cookies are stored and reused until invalid to minimize prompts
Browser cookie imports require the “Chrome Safe Storage” key from Keychain. Add CodexBar to the Access Control list in Keychain Access.app to prevent repeated prompts.

​
OAuth & API Tokens

  • Copilot: GitHub device flow
  • z.ai, Kimi K2, OpenRouter: API tokens stored in Keychain
  • Vertex AI: Google Cloud ADC (Application Default Credentials)

​
Local Probes

  • Antigravity: Language server HTTP endpoint
  • JetBrains AI: XML quota files from IDE configuration

​
Per-Provider Configuration

Each provider offers specific configuration options in Settings → Providers:
  • Enable/disable: Toggle providers on or off
  • Cookie source: Automatic browser import or manual cookie headers
  • Auth tokens: Store API keys securely in Keychain
  • Dashboard extras: Optional web scraping for additional metrics (Codex only)
  • Account switchers: Multiple token support for providers with account switching

​
Usage Metrics Tracked

Session Limits

5-hour or per-session usage windows

Weekly Limits

7-day rolling windows with reset countdowns

Credits

Prepaid credit balances and consumption rates

Rate Limits

Per-minute or per-hour request throttling

​
Status Monitoring

CodexBar polls status pages for supported providers and displays incident badges:
  • Statuspage.io integration: OpenAI (Codex), Anthropic (Claude), GitHub (Copilot), Cursor
  • Google Workspace: Gemini and Antigravity status
  • Factory Status: Droid/Factory incidents
  • Icon overlays: Visual indicators appear on menu bar icons during incidents

​
Local Cost Tracking

For Codex and Claude, CodexBar scans local JSONL logs to calculate token costs:
  • Codex: ~/.codex/sessions/**/*.jsonl
  • Claude: ~/.config/claude/projects/**/*.jsonl
  • Vertex AI: Claude logs filtered to Vertex AI-tagged entries
  • Time range: Last 30 days by default
  • Privacy-first: All parsing happens on-device
Use the bundled CLI for cost reporting: codexbar cost --provider codex or codexbar cost --provider claude

​
Adding New Providers

CodexBar is extensible. See the provider authoring guide to contribute support for additional AI services.

Menu Bar Interface

Understand the two-bar meter visualization

Privacy & Security

Learn how CodexBar protects your data

Build docs developers (and LLMs) love

Get started for freeTalk to us