DEV Community: Hojayfa Rahman

aswap: multi-account switching for the Antigravity CLI, with failover when quota runs out

Hojayfa Rahman — Fri, 02 Oct 2026 06:15:46 +0000

I use the Antigravity CLI (agy) with more than one Google account. Switching meant signing out and back in, and when an account ran out of quota in the middle of a long task, the task just stopped.

claude-swap solves exactly this for Claude Code, so I built the same thing for agy: antigravity-swap, command aswap.

hrmasss / antigravity-swap

Multi-account switcher for the Antigravity CLI (agy), with automatic rotation and quota-aware failover

antigravity-swap

Multi-account switcher for the Antigravity CLI (agy). Switch between Google accounts without signing out, let aswap switch for you before an account runs out, see every account's quota in one table, and run long print-mode tasks that carry on to the next account when one hits its limit.

Inspired by claude-swap by Onur Cetinkol, which does the same for Claude Code. The command set, the defaults and the auto-switch design come from it; see Credits.

Installation

Using uv (recommended)

uv tool install antigravity-swap

Using pipx

pipx install antigravity-swap

Standalone binary

No Python needed. Linux (x86_64, aarch64) and macOS (Apple silicon):

curl -fsSL https://raw.githubusercontent.com/hrmasss/antigravity-swap/main/install.sh | sh

Windows (PowerShell):

$d = "$env:LOCALAPPDATA\Programs\aswap"; New-Item -ItemType Directory -Force $d | Out-Null
irm https://github.com/hrmasss/antigravity-swap/releases/latest/download/aswap-windows-x86_64.exe -OutFile "

…

View on GitHub

Install

uv tool install antigravity-swap
# or: pipx install antigravity-swap

There are also standalone binaries for Linux, macOS and Windows on the releases page, and a one-line installer for Linux and macOS.

Add your accounts

aswap add

See every account at once

$ aswap list
Accounts:
  1: you@example.com [home] (active)
     ├ Gemini 5h:      23%   resets 13:42         in 4h 34m
     ├ Gemini 7d:       5%   resets Oct 8 18:23   in 6d 9h
     ├ Claude/GPT 5h:   0%   full
     └ Claude/GPT 7d:  33%   resets Oct 6 14:06   in 4d 4h
  2: work@example.com [work]
     ├ Gemini 5h:       0%   full
     ├ Gemini 7d:      44%   resets Oct 5 13:02   in 3d 3h
     ├ Claude/GPT 5h:   0%   full
     └ Claude/GPT 7d:   0%   full

Running instances:
  ● TUI    ~/code/app      1: you@example.com [home]   (2 sessions)

agy meters two pools separately, the Gemini models and the Claude/GPT models, each with a 5-hour and a weekly window. aswap list shows both, with reset times, and every agy session running on the machine with the account it is on.

Switch

aswap switch            # next account
aswap switch work       # by number, email or alias
aswap switch --strategy most-quota

After a switch, aswap runs agy models to prove the login works. That call lists models and spends no quota.

The part I actually wanted: keep a session going

aswap run

aswap run starts agy in your terminal and watches its log. When the account hits its quota, aswap:

holds that account out until the reset time agy reports,
picks the next account with the most quota left,
restarts agy on that account with the same conversation and a short "please continue" prompt.

You see agy restart and pick the task back up. For scripts and unattended runs, aswap exec -- -p "..." does the same for print mode.

Why a restart and not a swap in place?

claude-swap can change the account under a live Claude Code session, because Claude Code re-reads its credentials file. agy keeps its login in memory for the life of the process. I tested it: replacing the login file under a running agy changes nothing. So aswap restarts on the same conversation, which works because agy stores conversations as local files that any account can resume with --conversation <id>.

Two details mattered:

agy retries a quota error a couple of times before giving up. aswap only reacts to the final error that ends the turn, so a retry never kills a healthy session.
A fresh agy directory opens on the first-run theme picker. aswap carries over the "setup finished" flag, so a restart lands straight in the conversation.

Strategies

Strategy	Also called	Picks
`best`	`most-quota`	most quota left (default)
`rotate`	`round-robin`	next account by number
`next-available`		next account that still has quota
`consume-first`	`soonest-reset`	the account whose weekly quota resets soonest

There is also aswap auto, which watches the active account and switches before it runs out, with a threshold, hysteresis and a cooldown so two accounts near the limit never ping-pong. That design comes straight from claude-swap.

Platforms

On Linux each account gets its own agy directory, so several accounts run side by side. On Windows, agy keeps one login per machine in Credential Manager, so aswap switches that login, and briefly swaps a non-active account in to renew its token so its quota stays visible.

Credits

The command set, the defaults and the auto-switch design are borrowed from claude-swap by Onur Cetinkol. No code was copied; agy's login and session model needed its own implementation. MIT licensed, issues and PRs welcome.

GitHub: https://github.com/hrmasss/antigravity-swap
PyPI: https://pypi.org/project/antigravity-swap/

Why your React tournament bracket breaks in Safari (and a 4 KB pure-CSS fix)

Hojayfa Rahman — Sun, 31 May 2026 12:36:26 +0000

You build a tournament bracket with a popular React library. In Chrome it's perfect — neat columns, clean connector lines. Then you open it on an iPhone, or in Safari, or inside your Capacitor app… and every match is crammed into the top-left corner, stacked on top of the round headers.

If you've ever shipped a bracket to iOS, you've probably seen this exact bug. Here's why it happens — and a tiny library that fixes it for good.

The symptom

It looks fine everywhere Chromium runs (Chrome, Edge, Android WebView) and completely broken everywhere WebKit runs:

Safari (macOS and iOS)
iOS WKWebView
Capacitor / Cordova apps
Electron-on-WebKit

The matches don't just shift a little — they all render at coordinate (0,0) of the bracket, piling on top of each other and the headers.

The cause: SVG `<foreignObject>` in WebKit

Most React bracket libraries — @g-loot/react-tournament-brackets, react-tournament-bracket, and friends — render the bracket as an SVG and place each match's HTML inside a <foreignObject> positioned with x/y attributes.

WebKit has a long-standing bug: it ignores x, y, and transform on <foreignObject> and positions the content relative to the top-level <svg> instead of the foreignObject's own coordinates. Every match therefore collapses to the origin.

And there's no CSS escape hatch — x, y, and transform are all ignored on foreignObject in Safari, so you can't nudge the content back into place. I even tried patching a library to wrap each match in a <g transform="translate(x,y)"> instead of a nested <svg x y>; WebKit ignores ancestor transforms for foreignObject positioning too. The SVG approach is simply a dead end on WebKit.

The fix: don't use SVG at all

A bracket is really just columns of cards joined by connector lines — and both are expressible in plain CSS.

Here's the key insight. Put each round in a flex column where every match sits in an equal flex: 1 slot. Because each round has half the matches of the previous one, a match's slot spans exactly two feeder slots — so the two feeders land at 25% and 75% of that slot, and the match itself at 50%:

Round 1 slot   Round 2 slot
┌──────────┐
│ Match A  │──┐   25%
├──────────┤  ├─►┌──────────┐
│ Match B  │──┘  │ Winner   │   50%
└──────────┘     └──────────┘
┌──────────┐
│ Match C  │──┐   75%
│   ...    │

Draw the connector elbow with a few absolutely-positioned, bordered <div>s at those same 25 / 50 / 75% offsets, and the tree stays perfectly aligned at any height — with zero JavaScript measuring and no SVG. Because it's all flexbox and borders, it renders identically on Chromium, Firefox, and WebKit.

bracketkit

I packaged this up as bracketkit — a headless, pure-CSS tournament bracket for React:

🍏 Works in Safari / WebKit — no SVG, no foreignObject.
🧩 Headless — you render the match card; bracketkit owns layout + connectors. No theme objects, no design lock-in.
🪶 ~4 KB, zero dependencies — ESM + CJS + first-class TypeScript types.
⚡ SSR-safe — no measurement, correct on the first server render.
🎨 Style it any way — plain CSS, Tailwind, or a drop-in shadcn/ui component.

Quick start

npm i bracketkit

import { Bracket, type BracketRound } from "bracketkit"

type Match = { id: string; home: string; away: string; homeScore?: number; awayScore?: number }

const rounds: BracketRound<Match>[] = [
  {
    id: "sf",
    name: "Semi-finals",
    matches: [
      { id: "sf1", home: "Lions", away: "Bears", homeScore: 2, awayScore: 1 },
      { id: "sf2", home: "Hawks", away: "Wolves", homeScore: 0, awayScore: 3 },
    ],
  },
  { id: "f", name: "Final", matches: [{ id: "f1", home: "Lions", away: "Wolves" }] },
]

export function Playoffs() {
  return (
    <div style={{ overflowX: "auto", color: "#64748b" /* connector color */ }}>
      <Bracket
        rounds={rounds}
        renderRoundHeader={(round) => <h3>{round.name}</h3>}
        renderMatch={(m) => (
          <div className="match-card">
            <div>{m.home} — {m.homeScore ?? "–"}</div>
            <div>{m.away} — {m.awayScore ?? "–"}</div>
          </div>
        )}
      />
    </div>
  )
}

Theming

bracketkit ships no visual styling beyond layout. Connectors inherit currentColor and expose two CSS variables; every part has a data-* hook:

[data-bracket-root] {
  --bracket-connector-color: #64748b;
  --bracket-connector-width: 2px;
}
[data-bracket-match] { /* your card wrapper */ }

Prefer shadcn/ui?

npx shadcn@latest add https://hrmasss.github.io/bracketkit/r/bracket.json

You get a styled, batteries-included <Bracket> using your shadcn tokens — and you own the code.

Try it

🔗 Live demo & docs: https://hrmasss.github.io/bracketkit/
📦 npm: https://www.npmjs.com/package/bracketkit
⭐ GitHub: https://github.com/hrmasss/bracketkit

If you've fought the Safari foreignObject bug, I'd love to know whether this saves you the headache. Issues and PRs welcome.