CodexBar
github.com/steipete/codexbar- Category
- Developer Tools
- Rank
- No. 117Tools index
- Pricing
- Open Source
- Type
- APP
- Builder
- @steipete
- GitHub
- 21.1k stars
- Latest release
- v0.56.8
- Date
About
A macOS menu bar app that tracks your usage limits and quotas for AI coding tools like OpenAI Codex, Claude, Cursor, Copilot, and 15+ other providers. Shows real-time usage meters and reset countdowns without requiring separate logins.
What it does
CodexBar turns scattered provider accounting into one operational view. It gathers quota windows, credit balances, spending history, account details, and service incidents through each provider’s available authentication source. The menu app supports separate or merged provider icons, while the companion command line tool exposes the same data to terminals, scripts, dashboards, and a local web service.
Why it's ranked here
The breadth has real engineering depth, not just a long compatibility list. CodexBar handles provider-specific fallback chains, partial failures, reset windows, cost histories, and differing account scopes. Its command line output and guarded HTTP service make the collected data useful beyond the menu bar. The main reservation is inherent fragility: many integrations depend on browser sessions, local files, internal web endpoints, or external command output.
What's good
Authentication choices match how each service actually works. Providers can use API keys, OAuth credentials, device flows, signed-in command line tools, browser cookies, or local databases. Several fetchers preserve useful partial results when optional history or enrichment fails. Automation support is unusually practical: structured JSON, quota threshold checks, stable exit codes, atomic dashboard snapshots, caching, and row-level provider errors are all documented.
Tradeoffs
Setup varies sharply by provider. Some sources require API keys, manual cookies, another command line tool, or Full Disk Access for browser data. Web-backed integrations can break when private endpoints or session formats change. The desktop interface requires macOS 14 or newer, although the command line tool also supports Linux. Its built-in server has no TLS, CORS, or daemon mode, and remote binding requires explicit bearer-token and plain-HTTP acknowledgement.
How to use it well
It best suits developers juggling several AI subscriptions, especially when reset timing affects which service should handle the next long task. Keep only active providers enabled, prefer official API or command line credentials where available, and use JSON or quota guards in scripts and CI. Use the local server for a personal dashboard with careful network exposure. It does not replace provider billing consoles, guarantee every provider exposes complete cost data, or provide a general observability platform.
Technical notes+
README.md describes the Swift-based macOS 14+ menu application, provider toggles, status polling, cost scans, SQLite-backed Codex history, Homebrew installation, and cross-platform CLI tarballs. docs/cli.md documents a Commander-based CodexBarCLI built with Swift 6.2+, text and JSON output, quota guard exit codes, atomic dashboard snapshots, and an HTTP server with config-keyed caching, stale-response fallback, loopback host checks, bearer-token protection, and no TLS or CORS. Makefile exposes Swift build, lint, format, test, live-test, TTY-test, packaging, and release targets. Provider documents such as docs/kimi.md show the integration pattern: ordered credential resolution, API and browser fallbacks, explicit endpoint restrictions, snapshot mapping, and provider-specific tests.
Observed
- License
- MIT
- Primary language
- Swift
- Desktop platform
- macOS 14 or newer
- CLI platforms
- macOS and Linux, with glibc and static musl Linux tarballs
- Install surfaces
- GitHub Releases, Homebrew cask, Homebrew formula, Arch Linux AUR, and release tarballs
- Interfaces
- macOS menu bar application, command line interface, JSON snapshots, and local HTTP service
- Build tooling
- Swift Package Manager targets exposed through Make
Read from README.md, Makefile, docs/cli.md, docs/amp.md, docs/poe.md, docs/TODO.md, docs/crof.md, docs/grok.md, docs/groq.md, docs/icon.md, docs/kilo.md, docs/kimi.md, docs/kiro.md, docs/mimo.md, docs/aiand.md.
What it can do
Track AI coding tool usage quotas
AI provider accounts (OpenAI Codex, Claude, Cursor, Copilot, etc.) → Real-time usage statistics and remaining quota limits
Display usage meters in menu bar
Current usage data from multiple AI providers → Visual usage meters showing consumption levels
Monitor session usage tracking
Active coding session data from AI tools → Current session usage statistics
Monitor weekly usage tracking
Historical usage data over weekly periods → Weekly usage summaries and trends
Show quota reset countdowns
Provider quota reset schedules and current time → Countdown timers until quota limits reset
Aggregate usage across 15+ AI providers
Usage data from multiple AI coding services → Unified dashboard of all provider usage statistics
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.