Live usage limits for 17 AI providers, right in your menu bar

Every AI coding tool rations you differently. Claude Code has a 5-hour session window and a weekly ceiling. Codex has its own 5-hour primary window. Cursor mixes an included plan with bonus and on-demand spend. Gemini CLI waterfalls you from Pro to Flash to Flash Lite as quotas drain. None of them tell you where you stand until you hit the wall mid-task.

AgentHarbor's analytics subsystem exists to make that visible. The tray icon shows the most relevant live metric for your active provider — session percentage for Claude Code, total spend for Cursor, quota for Gemini — refreshed about every 120 seconds. The popover breaks it down per provider with quota bars and reset times.

The limit-state ladder

Under the hood, every provider is tracked on a single ladder: Unauthenticated → ApiDisabled → BillablePaused → SubscriptionIssue → RateLimited → Reached → Approaching → Healthy. Whenever a provider transitions, AgentHarbor fires a native notification — so "approaching your session limit" arrives while you can still plan around it, not after your agent stalls.

The tray icon itself switches to a red variant whenever the active provider is in any non-healthy state, which makes a glance at the menu bar enough.

What your subscription is actually worth

The cost engine computes per-model API-equivalent costs — Claude Opus, Sonnet, and Haiku, plus Codex models — token-deduplicated across session windows. That turns "I used 62% of my session" into "this month's usage would have cost $X via API," which is the number that actually matters when you're deciding between plans.

Seventeen providers and counting

Today AgentHarbor tracks Claude Code, Cursor, Codex (OpenAI), Gemini CLI, Claude Desktop, GitHub Copilot, Kiro, Vertex AI, JetBrains AI, OpenRouter, Augment, Amp, Droid (Factory), Kimi, Kimi K2, z.ai, and Windsurf. Most auto-detect your existing CLI or IDE auth; the rest connect with an API key, token, or OAuth device flow.

Everything runs locally. The only outbound calls are to each provider's official API using your own tokens — no telemetry, no middleman, nothing leaves your machine.