<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Linux - Tag - arleo.eu</title><link>https://www.arleo.eu/en/tags/linux/</link><description>Linux - Tag - arleo.eu</description><generator>Hugo -- gohugo.io</generator><language>en</language><lastBuildDate>Sat, 09 May 2026 13:02:29 +0200</lastBuildDate><atom:link href="https://www.arleo.eu/en/tags/linux/" rel="self" type="application/rss+xml"/><item><title>systemd hardening: taking a Python service from 9.6 to 1.7</title><link>https://www.arleo.eu/en/posts/hardening-systemd-mcp/</link><pubDate>Sat, 09 May 2026 13:02:29 +0200</pubDate><author>Jmr</author><guid>https://www.arleo.eu/en/posts/hardening-systemd-mcp/</guid><description><![CDATA[<div class="featured-image">
                <img src="/images/hardening-systemd-mcp-featured.jpg" referrerpolicy="no-referrer">
            </div><h2 id="tldr">TL;DR</h2>
<p><code>systemd-analyze security</code> is an underused tool. It scans your unit files and computes an exposure score from <strong>0.0 (UNSAFE)</strong> to <strong>10.0 (PERFECT)</strong>. Custom Python services often score around <strong>9.6</strong> by default — that&rsquo;s bad.</p>
<p>I took my <code>hugo-mcp</code> service (FastAPI exposing 7 MCP tools) from <strong>9.6 → 1.7</strong> without breaking a single feature. Here are the directives that actually matter, and the ones that are traps.</p>
<h2 id="initial-score">Initial score</h2>]]></description></item></channel></rss>