HelloAssembly The smallest possible complete Windows application
github.com
August 25, 2026
4 min read
46/100
Summary
PlummersSoftwareLLC’s HelloAssembly repository contains x86 assembly implementations of a minimal complete Windows application. The project targets an uncompressed executable that runs a Windows message loop; provides a title bar with functional minimize, maximize, and close controls; includes the corresponding system menu; and paints a background plus centered text at least as large as “Dave’s Tiny App.” The repository includes several variants: TinyOriginal, an optimized original implementation; Lasse, which uses shell-coding tactics; and Theron, which uses a manually written Portable Executable header. As of March 11, 2023, the listed smallest working builds were 1,104 bytes for Lasse built with MASM32, 818 bytes for Lasse built with Crinkler, 540 bytes for TinyOriginal built with Crinkler, and 383 bytes for Theron built with Yasm. Lasse can be built with MASM32 11.0, while Theron requires Yasm. Crinkler, a linker designed for executables of a few kilobytes, requires the Windows SDK; version 10.0.20348.0 produced the smallest stated results. The repository warns that aggressive size optimizations in Lasse and Crinkler may resemble malware techniques and trigger antivirus detections. The maintainers say the repository contains no malicious code, while warning they cannot verify binaries or code obtained from other sources.
Key Takeaways
- HelloAssembly implements a complete Windows GUI program in x86 assembly while minimizing executable size without compression.
- The Theron implementation, built with Yasm and a manually written PE header, was listed at 383 bytes as of March 11, 2023.
- Crinkler is a Windows compressing linker used for tiny executables; the repository states that Windows SDK 10.0.20348.0 achieved its best results.
- Aggressive executable-size optimizations can cause antivirus software to flag locally built binaries as suspected malware.
What the discussion said
The thread was overwhelmingly about tiny Windows executables, what counts as a complete application, and the author’s reputation rather than AI. AI surfaced only in two side remarks. One reader pointed to a prior personal project: an AI agent reportedly surpassed its creator’s high score in the arcade game Tempest. That was offered as evidence that the developer has done technically interesting work, but nobody examined the system’s training, evaluation, or broader significance. A second reader was optimistic that coding agents could make native assembly more practical: rather than hand-porting a small application to every target, they imagined asking Codex to translate it across hosts. That is the thread’s only substantive AI-application claim, and it rests on AI reducing the cost of low-level portability. No one challenged that forecast directly, nor did commenters discuss whether generated assembly is reliable, secure, performant, or maintainable. The sharper arguments about binary size, system libraries, and Windows APIs were not AI-related and do not establish sentiment about AI. Consequently, the AI reaction is lightly positive but extremely thin: readers showed curiosity about AI beating a human game score and enthusiasm for agent-assisted cross-platform work, without enough discussion to support strong conclusions.
Community Sentiment
Positives
- An AI agent reportedly exceeded its creator’s Tempest high score, a compact example of machine play surpassing an expert on a tightly defined task.
- One commenter sees coding agents as a route to portable native assembly, turning cross-platform rewrites into an automated translation job rather than a manual slog.