CodeEdge
CodeEdge
Code to the edge.
Codex, Claude, Cursor, Gemini, Copilot, Kiro and 100+ more,pinned to the edge of your screen.Remaining quota, reset times, today’s tokens and estimated spend,without opening a single dashboard.
macOS 26.5 or later · Apple silicon and Intel
Is CodeEdge for you?
Every AI service shows its limits somewhere different: a settings page, a CLI command, a billing console, a usage tab three clicks deep. When you pay for several of them, checking what’s left becomes a chore, and you usually find out at the worst moment, when a request is refused.
CodeEdge keeps all of it in a slim rail at the edge of your screen.
It’s for you if you use more than one AI tool or plan. If you use one service and its own usage page is enough, you don’t need it.
What it does
- Rail
- One ring per provider, filled to the share of quota that’s left. It floats at either edge of the screen, stays there across Spaces and remembers its position.
- Details
- Click a ring for the plan, every quota window (session, 5-hour, weekly, monthly, credits, balance), how much of each is used, and when it resets.
- Spend
- Today’s and the last 30 days’ tokens from the local session logs of Codex, Claude Code, Pi and DeepSeek, priced with each model’s rates.
- Real data
- Quotas come from each provider’s own source: its API, a local account, its CLI or a browser session. Nothing is guessed from chat text or logs.
- Native
- SwiftUI and AppKit. System, dark and light appearance, English and Simplified Chinese, launch at login and in-app updates.
Supported providers
- Codex
- Claude
- Cursor
- Gemini
- Copilot
- Kiro
- Windsurf
- Amp
- Zed
- Warp
- OpenAI
- OpenRouter
- DeepSeek
- Mistral
- Kimi Code
- z.ai / GLM
- MiniMax
- Ollama
- Claude Platform
- Antigravity
- xKiro
- Augment
- Devin
- Droid
- JetBrains AI
- Qoder
- Trae CN
- CodeRabbit
- Codebuff
- Command Code
- ClinePass
- OpenCode
- OpenCode Go
- Kilo
- Muse Code
- Pi
- WorkBuddy
- Raycast
- Azure OpenAI
- AWS Bedrock
- Vertex AI
- Grok
- xAI
- Groq
- Perplexity
- Moonshot
- Alibaba
- Alibaba Token Plan
- Qwen Cloud
- Doubao
- Xiaomi MiMo
- StepFun
- LongCat
- Fireworks
- DeepInfra
- Hugging Face
- Replicate
- fal
- Venice
- Chutes
- Synthetic
- Poe
- T3 Chat
- Manus
- Notion AI
- Sakana AI
- Abacus AI
- ElevenLabs
- Deepgram
- Soniox
- Tavily
- Exa
- Linkup
- Parallel
- Vercel AI Gateway
- v0
- LiteLLM
- LLM Proxy
- ZenMux
- ClawRouter
- Wayfinder
- Bifrost
- sub2api
- Neuralwatt
- Atlas Cloud
- Langdock
- IBM Bob
- GitKraken AI
- Nous Portal
- Charm Hyper
- Chatbox AI
- Future AGI
- Munito
- OpenDesign
- abliteration.ai
- SiliconFlow
- Tripo
Each provider is read the way its service allows: a local account, a CLI, an API key, a browser session or a custom endpoint. Provider logic follows CodexBar.
Privacy
- Keys
- Stored in the macOS Keychain.
- Local
- Usage caches, preferences and the rail’s position stay on your Mac.
- Logs
- Read only for token and cost totals. Chat and code text are never parsed into quotas.
- Network
- Only each provider’s own data source, plus GitHub Releases for updates.
Quick start
- InstallDownload the DMG and drag CodeEdge into Applications.
- Enable providersIn Settings → Providers, turn on the services you use.
- GlanceRings appear in the rail as providers report data. Click one for details.
CodeEdge isn’t signed with an Apple Developer ID or notarised yet, so macOS may block the first launch. Open System Settings → Privacy & Security and choose to open it anyway.
Made by
Seaony
Next year brings its own snow.
Singapore
A full-stack developer who makes products.What I build changes from one project to the next,but I care about the same thing every time:every detail has been looked after, and the people using it can feel it.
Other projects
mameya.app