timharris707

Modeldeck — Development skill for Claude Code

Development community

How to install Modeldeck

This entry records only its repository, not the path inside it, so there is no exact command to give. Open timharris707/modeldeck and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Modeldeck does

macOS menu bar app that tracks usage limits across all your Claude Code and Codex CLI accounts — live "% left" meters per rate-limit window, reset countdowns, low-capacity alerts, and one-click account switching. Local-first: no cloud, no telemetry, credentials never touched. Native SwiftUI + a localhost Node daemon.

Alternatives in Development

Sprint Plan — Plan sprints with capacity, story selection, and risk mapping 7.8k ★
Claude Swap — Switch between multiple Claude Code accounts, with automatic rate-limit rotation, usage dashboard, and paralle 2k ★
Plan Sprint — Plan the upcoming sprint — epic breakdown, user story decomposition, RICE prioritization, dependency check, an 73 ★

README

ModelDeck

**One menu bar icon that always knows how much Claude Code and Codex you have left.**

A native macOS menu bar app + local daemon that tracks usage limits across all of your Claude Code and Codex CLI accounts — live "% left" meters, reset countdowns, and one-click account switching.

**Local-first. No cloud backend. No telemetry. Your provider credentials are never copied, stored, or transmitted by ModelDeck.** The only secrets ModelDeck creates are its own local Keychain items — the token that guards the daemon's API and, if you use the managed proxy, a client key per profile — and they contain nothing of yours. The only optional outbound calls of ModelDeck's own are the daily update check (reads this repository's public releases feed) and the update download you approve (fetches the release asset) — update checks are off unless you enable them.

[Download](https://github.com/timharris707/modeldeck/releases) · [modeldeck.ai](https://modeldeck.ai)

![Platform](https://img.shields.io/badge/platform-macOS%2014%2B-blue) ![Swift](https://img.shields.io/badge/app-Swift%20%2B%20SwiftUI-F05138) ![Node](https://img.shields.io/badge/daemon-Node.js%2024%2B-339933) ![Local-first](https://img.shields.io/badge/data-100%25%20local-success) ![License](https://img.shields.io/badge/license-PolyForm%20Noncommercial-blue)

![The ModelDeck deck popover — every account and limit window at a glance](docs/images/deck-popover.png)

*All screenshots show a demo-seeded instance with placeholder accounts.*

Why

If you run AI coding agents seriously, you probably don't have *one* account — you have a work Claude Max plan, a personal Pro plan, a Codex subscription, maybe a spare for side projects. Each one has its own 5-hour session window, weekly limit, and model-scoped caps, all resetting on different clocks.

Today the only way to know where you stand is to interrupt what you're doing and check each account by hand — usually right after an