MashuPack

<p>nice and simple utility, I've been using <a href="https://peteretelej.github.io/tree/" target="_blank" rel="nofollow noopener noreferrer"><strong>tree</strong></a> for this, but Mashupack is a nice visual way to solve the same problem.</p>

<blockquote><p>This solves a very real Claude/ChatGPT workflow pain. Simple idea, high utility = dangerous combo 🔥</p></blockquote><p></p>

<p>Do you keep file paths and a table of contents so the model can still reference where things came from?</p>

I built MashuPack because I had a split workflow. In the terminal, I already had good tools for navigating codebases and working with agents in real time. But I still do a lot of long-form planning, debugging, review, and codebase understanding in conversational web UIs like ChatGPT and Claude. That workflow kept breaking on context. File-count limits, upload limits, and inconsistent format support made it annoying to get the right slice of a codebase into the model. The one format that always seemed to work was a single text file. So I started manually packing up small collections of source files whenever I wanted to discuss a subsystem, plan a refactor, or ask high-level questions about a repository. That got repetitive fast. MashuPack came from wanting a better interface for that exact job: select the context you actually want, compile it into one clean file, and stay in control of what gets included. I’d especially love feedback from people who use ChatGPT or Claude in the browser for software planning, debugging, or understanding unfamiliar codebases.