From 410d48da574477d59ecad038986d9de786356a0a Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 29 Jun 2026 14:41:16 +0000 Subject: [PATCH] docs: center all media assets in README Co-Authored-By: Claude Opus 4.8 Claude-Session: https://claude.ai/code/session_01FMyhTMz8kqFP1TjZKyqbZw --- README.md | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 4b5e3ac..178b100 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,12 @@ +
Screenshot 2026-04-01 at 20 38 16 +
-[šŸŽ„Explanation Video](https://github.com/user-attachments/assets/230f789a-5d51-4eec-8848-81f96c742863) +

+šŸŽ„Explanation Video +

Desktest is a general computer use CLI for automated end-to-end virtualised testing of desktop applications using LLM-powered agents. Spins up a disposable šŸ³ Docker container (Linux), [Tart VM (macOS)](https://tart.run/), or QEMU/KVM VM (Windows) with a desktop environment, deploys any apps, and runs a computer-use agent that interacts with it based on your prompt. Built with coding agents in mind as first-class citizen users of `desktest`. @@ -25,7 +29,9 @@ Install the desktest CLI by running `curl -fsSL https://raw.githubusercontent.co ``` +
How Desktest works +
## Getting Started @@ -45,7 +51,9 @@ New to desktest? Follow the [Getting Started guide](GETTING_STARTED.md) for a wa Desktest uses the same agent harness as the [OSWorld](https://os-world.github.io/) benchmark for evaluating multimodal agents on real-world computer tasks. The leaderboard below tracks which models perform best, [updated weekly](https://github.com/Edison-Watch/osworld-leaderboard-updater): -[![OSWorld Leaderboard](https://raw.githubusercontent.com/Edison-Watch/osworld-leaderboard-updater/main/assets/osworld-leaderboard.svg)](https://github.com/Edison-Watch/osworld-leaderboard-updater) +

+OSWorld Leaderboard +

## Use Cases