<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Juanmidev</title>
    <description>The latest articles on DEV Community by Juanmidev (@juanmidev).</description>
    <link>https://dev.to/juanmidev</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4256407%2Fab04b4c1-6284-4186-83b2-6114b1ed5ada.png</url>
      <title>DEV Community: Juanmidev</title>
      <link>https://dev.to/juanmidev</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/juanmidev"/>
    <language>en</language>
    <item>
      <title>I Built repo-pack: A Single-Binary Rust CLI &amp; MCP Server to Pack Repos for LLMs</title>
      <dc:creator>Juanmidev</dc:creator>
      <pubDate>Fri, 20 Nov 2026 22:22:43 +0000</pubDate>
      <link>https://dev.to/juanmidev/i-built-repo-pack-a-single-binary-rust-cli-mcp-server-to-pack-repos-for-llms-4pfm</link>
      <guid>https://dev.to/juanmidev/i-built-repo-pack-a-single-binary-rust-cli-mcp-server-to-pack-repos-for-llms-4pfm</guid>
      <description>&lt;p&gt;When working with LLMs on real codebases, feeding the right context without blowing up your token budget (or waiting seconds for heavy runtimes to boot) is half the battle.&lt;/p&gt;

&lt;p&gt;Tools like Repomix are great, but I wanted something that shipped as a &lt;strong&gt;single native binary&lt;/strong&gt; with zero Node/Python runtime dependencies, ran in a few milliseconds offline, and could strip function implementations at the AST level when I only needed architectural context.&lt;/p&gt;

&lt;p&gt;So I built &lt;strong&gt;&lt;a href="https://github.com/Juanmidev1/repo-pack" rel="noopener noreferrer"&gt;repo-pack&lt;/a&gt;&lt;/strong&gt; in Rust.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F64h0hnkd8tku5huo4uvu.gif" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F64h0hnkd8tku5huo4uvu.gif" alt=" " width="800" height="596"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  What makes it different?
&lt;/h2&gt;

&lt;h3&gt;
  
  
  1. Single binary, zero runtime overhead
&lt;/h3&gt;

&lt;p&gt;&lt;code&gt;repo-pack&lt;/code&gt; compiles down to a single static binary (~8 MB with Tree-sitter grammars included, or &lt;strong&gt;under 3 MB&lt;/strong&gt; with &lt;code&gt;--no-default-features&lt;/code&gt;). It walks your project respecting &lt;code&gt;.gitignore&lt;/code&gt; and &lt;code&gt;.ignore&lt;/code&gt; (even outside git repos), skips binaries and lockfiles automatically, and outputs structured XML or Markdown in milliseconds.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. AST Skeleton Mode (&lt;code&gt;--skeleton&lt;/code&gt; &amp;amp; &lt;code&gt;--focus&lt;/code&gt;)
&lt;/h3&gt;

&lt;p&gt;Dumping an entire repository raw wastes massive amounts of context on function internals you don't care about.&lt;/p&gt;

&lt;p&gt;With &lt;code&gt;--skeleton&lt;/code&gt;, &lt;code&gt;repo-pack&lt;/code&gt; parses Rust, TypeScript/TSX, JavaScript/JSX, and Python using &lt;strong&gt;Tree-sitter&lt;/strong&gt;, replacing function and method bodies with &lt;code&gt;{ ... }&lt;/code&gt; (or &lt;code&gt;...&lt;/code&gt; in Python while preserving docstrings). Imports, types, structs, classes, and signatures stay intact.&lt;/p&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;
bash
# Skeletonize the entire repository for high-level architectural context
repo-pack . --skeleton

# Keep full code only for auth files, skeletonize everything else
repo-pack . --focus "src/auth/*"
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

</description>
      <category>ai</category>
      <category>rust</category>
      <category>mcp</category>
      <category>tokens</category>
    </item>
  </channel>
</rss>
