# Abralo

> A lightweight native desktop app for running several coding agents side by
> side, one per project folder, in one window that's built to read at a glance
> instead of scrolling a terminal. Drives the official `claude` and `codex` CLIs
> with your own Claude or OpenAI subscription. Free and unlimited. macOS,
> Windows, and Linux.

Abralo is for developers who run multiple coding agents at the same time and
want to see, at a glance, which ones are working, which need an answer, and how
much of their usage limit they're burning. It is not a cloud service and not an
LLM of its own. It runs the user's own local `claude` and `codex` CLIs.
Conversations, code, and prompts stay between the user and their model provider;
Abralo never sees them.

## Product

- [Home / overview](https://abralo.com/): what Abralo is, the readability and usage-monitoring story, and the download.
- [Abralo vs. Claude Code in a terminal](https://abralo.com/compare): honest side-by-side comparison, measured install footprint, and where a terminal is still the better tool.
- [Plain-text (Markdown) version of the home page](https://abralo.com/index.md): the same content as clean Markdown for LLM context.

## Download

- [Download for Windows](https://abralo.com/download/windows): Windows 10 or 11, 64-bit.
- [Download for macOS](https://abralo.com/download/mac): macOS 10.15+ , Apple Silicon or Intel. Signed and notarized.
- [Download for Linux](https://abralo.com/download/linux): 64-bit Linux, 2022 or newer (Ubuntu 22.04+, Debian 12+, Fedora 36+).
- [Releases & changelog](https://github.com/cw12574/abralo-app/releases): version history and signed installer downloads for every platform (release binaries only; source is private).

## Key facts

- Pricing: free and unlimited (no agent cap, no account, no time limit).
- Agents supported: Claude Code and Codex, chosen per column; a column can be switched to the other agent mid-conversation without losing context.
- Requirements: an existing Claude or OpenAI login (Abralo runs the official `claude` and `codex` binaries locally).
- Usage gauge: sparklines and session predictions cover Claude Code; Codex exposes no equivalent, so less is shown for those columns.
- Privacy: your conversations, code, and prompts are only ever between you and your model provider; anonymous, opt-out feature analytics only.
- Native: a real native build per platform, not one web page in three wrappers; the whole app is only a few megabytes.
- Contact: chris@abralo.com