<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[JavaScript in Plain English - Medium]]></title>
        <description><![CDATA[New JavaScript and Web Development content every day. Follow to join our 3.5M+ monthly readers. - Medium]]></description>
        <link>https://javascript.plainenglish.io?source=rss----4b3a1ed4f11c---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>JavaScript in Plain English - Medium</title>
            <link>https://javascript.plainenglish.io?source=rss----4b3a1ed4f11c---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 03:48:06 GMT</lastBuildDate>
        <atom:link href="https://javascript.plainenglish.io/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com" rel="hub"/>
        <item>
            <title><![CDATA[How to Build a VS Code-Style Command Palette with React and TypeScript]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/how-to-build-a-vs-code-style-command-palette-with-react-and-typescript-3b835d7116a0?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1672/1*Ydm7RaPOWS0JjOvzavCq-Q.png" width="1672"></a></p><p class="medium-feed-snippet">Add searchable commands, keyboard shortcuts, and arrow-key navigation to your React app &#x2014; without installing a UI library.</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/how-to-build-a-vs-code-style-command-palette-with-react-and-typescript-3b835d7116a0?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/how-to-build-a-vs-code-style-command-palette-with-react-and-typescript-3b835d7116a0?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/3b835d7116a0</guid>
            <category><![CDATA[react]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[typescript]]></category>
            <category><![CDATA[frontend-development]]></category>
            <dc:creator><![CDATA[Bestte]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 17:56:36 GMT</pubDate>
            <atom:updated>2026-11-17T17:56:34.836Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[I Built 3 JavaScript Side Projects That Now Pay My Rent]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/i-built-3-javascript-side-projects-that-now-pay-my-rent-9485dfef9675?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/2400/1*ZpBt0xcQmNj1OH3pMwK7Fw.png" width="2400"></a></p><p class="medium-feed-snippet">Here&#x2019;s exactly how a few nights of coding turned into a steady monthly income stream &#x2014; and the automation tricks that made it possible</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/i-built-3-javascript-side-projects-that-now-pay-my-rent-9485dfef9675?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/i-built-3-javascript-side-projects-that-now-pay-my-rent-9485dfef9675?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/9485dfef9675</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[freelancing]]></category>
            <category><![CDATA[side-project]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[web-development]]></category>
            <dc:creator><![CDATA[Mohsin Abro]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 17:45:25 GMT</pubDate>
            <atom:updated>2026-11-17T17:45:24.174Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[NPM Workspaces: Why Does My Monorepo Build Work Locally but Break in CI?]]></title>
            <link>https://javascript.plainenglish.io/npm-workspaces-why-does-my-monorepo-build-work-locally-but-break-in-ci-49bd8370b3c1?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/49bd8370b3c1</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Rajat]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 15:46:51 GMT</pubDate>
            <atom:updated>2026-11-17T15:46:50.337Z</atom:updated>
            <content:encoded><![CDATA[<h4>One root install, one lockfile, one set of versions, and the build-order trap nobody warns you about</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j85YfYQ7vQ_sTKh5tdLEqg.jpeg" /></figure><p>It’s 6:40 pm. Your pull request is green on your laptop. You push, CI goes red, and the error says Cannot find module &#39;@acme/core&#39;. You didn&#39;t touch that import. Nobody did. You re-run the pipeline anyway, because hope is a strategy, and it fails again.</p><p>If you’ve got two or three packages sharing code in one repo, you’ve probably met some version of this. NPM workspaces are the built-in way to keep those packages consistent: one install, one lockfile, shared dependency versions, and one command that runs a script everywhere. But the feature has sharp edges that most tutorials skip.</p><p>Here’s the open question for this article: <strong>why did the build pass on your machine and fail on a clean checkout?</strong> We’ll answer it before the end.</p><p>If you like practical monorepo write-ups like this, follow me here on Medium. I post one every week.</p><p>By the end you’ll know how to set up workspaces, run scripts recursively, filter to the packages you care about, fake parallel runs (npm doesn’t do them natively), and avoid the hoisting trap that hides missing dependencies.</p><h3>What are NPM workspaces, really?</h3><p>Workspaces let one root package.json manage several packages in the same repository. They landed in npm 7, so any modern Node install has them. Check yours:</p><pre>npm -v</pre><p>Before you rely on any flag in this article, skim the <a href="https://docs.npmjs.com/cli/using-npm/workspaces">official npm workspaces docs</a> for your version. The CLI is still moving.</p><blockquote><em>Before we get into the examples, a quick note: The code snippets provided here are meant purely for understanding the concept. Always refer to the official documentation for the most current API and syntax.</em></blockquote><p>We’ll build one small repo and grow it as we go. Two packages: @acme/core (a tiny shared library) and @acme/app (a consumer). Both are TypeScript.</p><h3>Quick version: the minimum setup</h3><p>Create this layout:</p><pre>acme/<br>  package.json<br>  tsconfig.base.json<br>  packages/<br>    core/<br>      package.json<br>      tsconfig.json<br>      src/index.ts<br>    app/<br>      package.json<br>      tsconfig.json<br>      src/index.ts</pre><pre>// acme/package.json<br>{<br>  &quot;name&quot;: &quot;acme&quot;,<br>  &quot;version&quot;: &quot;1.0.0&quot;,<br>  &quot;private&quot;: true,<br>  &quot;workspaces&quot;: [&quot;packages/*&quot;],<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;npm run build --workspaces&quot;<br>  },<br>  &quot;devDependencies&quot;: {<br>    &quot;@types/node&quot;: &quot;^22.0.0&quot;,<br>    &quot;typescript&quot;: &quot;^5.6.0&quot;<br>  }<br>}</pre><p>(Strip the // comment line if you paste these into real JSON files. I&#39;m using it as the file name label.)</p><p>Two things matter here. &quot;private&quot;: true stops you from accidentally publishing the root. And TypeScript is installed once, at the root, so every package uses the same compiler version. That&#39;s the &quot;make things the same&quot; part: one version of the tool, everywhere.</p><pre>// acme/tsconfig.base.json<br>{<br>  &quot;compilerOptions&quot;: {<br>    &quot;target&quot;: &quot;ES2022&quot;,<br>    &quot;module&quot;: &quot;commonjs&quot;,<br>    &quot;strict&quot;: true,<br>    &quot;declaration&quot;: true,<br>    &quot;esModuleInterop&quot;: true,<br>    &quot;skipLibCheck&quot;: true<br>  }<br>}</pre><pre>// acme/packages/core/package.json<br>{<br>  &quot;name&quot;: &quot;@acme/core&quot;,<br>  &quot;version&quot;: &quot;1.0.0&quot;,<br>  &quot;main&quot;: &quot;dist/index.js&quot;,<br>  &quot;types&quot;: &quot;dist/index.d.ts&quot;,<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;tsc&quot;<br>  }<br>}</pre><pre>// acme/packages/core/tsconfig.json<br>{<br>  &quot;extends&quot;: &quot;../../tsconfig.base.json&quot;,<br>  &quot;compilerOptions&quot;: {<br>    &quot;rootDir&quot;: &quot;src&quot;,<br>    &quot;outDir&quot;: &quot;dist&quot;<br>  },<br>  &quot;include&quot;: [&quot;src&quot;]<br>}</pre><pre>// acme/packages/core/src/index.ts<br>export function formatPrice(cents: number): string {<br>  return `$${(cents / 100).toFixed(2)}`;<br>}</pre><p>Now the consumer:</p><pre>// acme/packages/app/package.json<br>{<br>  &quot;name&quot;: &quot;@acme/app&quot;,<br>  &quot;version&quot;: &quot;1.0.0&quot;,<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;tsc&quot;,<br>    &quot;start&quot;: &quot;node dist/index.js&quot;<br>  },<br>  &quot;dependencies&quot;: {<br>    &quot;@acme/core&quot;: &quot;*&quot;<br>  }<br>}</pre><pre>// acme/packages/app/tsconfig.json<br>{<br>  &quot;extends&quot;: &quot;../../tsconfig.base.json&quot;,<br>  &quot;compilerOptions&quot;: {<br>    &quot;rootDir&quot;: &quot;src&quot;,<br>    &quot;outDir&quot;: &quot;dist&quot;<br>  },<br>  &quot;include&quot;: [&quot;src&quot;]<br>}</pre><pre>// acme/packages/app/src/index.ts<br>import { formatPrice } from &quot;@acme/core&quot;;</pre><pre>console.log(formatPrice(1999));</pre><p>Run one install from the root:</p><pre>cd acme<br>npm install<br>ls -l node_modules/@acme</pre><p>Expected output (trimmed):</p><pre>core -&gt; ../../packages/core<br>app -&gt; ../../packages/app</pre><p>Those are symlinks. Your workspace packages show up in the root node_modules as links to the real folders, so import &quot;@acme/core&quot; resolves to your local source instead of the registry. Edit core, and the app sees it. No npm link dance.</p><p><strong>Real-world version.</strong> Notice the &quot;*&quot; range for @acme/core. npm sees that a local workspace satisfies it and links the folder instead of fetching anything. If you pin a version the workspace doesn&#39;t match, say ^2.0.0 while core is at 1.0.0, npm will go to the registry and install a copy from there. That&#39;s a quiet source of &quot;why is my local change ignored&quot; bugs.</p><p><strong>Under the hood.</strong> There’s a single package-lock.json at the root. Workspaces are treated as link nodes in one dependency tree, which npm&#39;s tree builder (Arborist) dedupes and hoists to the root node_modules where it can. One tree, one lockfile, one answer to &quot;which version of this is installed?&quot;</p><p>So far everything looks smooth. What happens when you run the build for the whole repo?</p><h3>What does a recursive call actually do?</h3><p>The answer to “run this script in every package” is the --workspaces flag, or -ws for short.</p><h3>Quick version</h3><pre>npm run build --workspaces</pre><p>npm walks each workspace and runs its build script. If some packages lack the script, add --if-present so npm skips them instead of failing:</p><pre>npm run build -ws --if-present</pre><p>Expected output (trimmed):</p><pre>&gt; @acme/app@1.0.0 build<br>&gt; tsc</pre><pre>&gt; @acme/core@1.0.0 build<br>&gt; tsc</pre><p>Look at that order. App built before core. Keep that in mind.</p><h3>Real-world version: the wrong version that still runs</h3><p>Here’s the trap. On your laptop, you already built core yesterday, so packages/core/dist exists. App&#39;s tsc finds the old dist/index.d.ts, compiles happily, and the whole thing looks fine.</p><p>On a clean CI checkout, dist doesn&#39;t exist. App builds first (workspaces from a glob like packages/* typically resolve alphabetically, and app comes before core), and TypeScript can&#39;t find the types:</p><pre>src/index.ts(1,28): error TS2307: Cannot find module &#39;@acme/core&#39; or its corresponding type declarations.</pre><p>That’s the answer to our open question. The recursive flag runs workspaces in the order npm lists them. It does <strong>not</strong> read your dependency graph. Locally, a stale dist folder was covering for the missing ordering.</p><p>The fix is to be explicit about order in the root script:</p><pre>// acme/package.json (scripts section)<br>{<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;npm run build -w @acme/core &amp;&amp; npm run build -w @acme/app&quot;<br>  }<br>}</pre><p>Now clean checkouts and warm laptops behave the same. To prove it to yourself, delete both dist folders and run npm run build from the root. It should succeed every time.</p><p>Have you ever had a build that only worked because of leftover output on your disk? Hold that thought, there’s a cousin of it later in the article.</p><p><strong>Under the hood.</strong> --workspaces is a config switch that tells the command to run in the context of all configured workspaces. Commands that touch the dependency tree (install, update) link workspaces into node_modules. Commands that run things (test, exec, publish) go to the root project unless you pass a workspace flag. That&#39;s why a bare npm test at the root doesn&#39;t run your packages&#39; tests. It runs the root&#39;s.</p><p>The root script solved ordering, but typing -w @acme/core all day gets old. Is there a better way to aim at one package?</p><h3>How do you filter to just one workspace?</h3><p>NPM doesn’t have a separate “filter” command. The -w (or --workspace) flag is the filter.</p><h3>Quick version</h3><pre>npm run build -w @acme/core<br>npm run build -w packages/core<br>npm run build -w core</pre><p>All three target the same package. The docs say valid values are a workspace name, a path to a workspace directory, or a path to a parent directory, which selects every workspace inside it. The short folder name works in practice too, but I’d use the package name in scripts because it survives folder renames.</p><p>Pass -w more than once to pick several:</p><pre>npm run build -w @acme/core -w @acme/app</pre><h3>Real-world version</h3><p>Adding dependencies is where -w earns its keep:</p><pre>npm install dayjs -w @acme/core<br>npm uninstall dayjs -w @acme/core</pre><p>The dependency lands in core’s package.json, and the root lockfile updates. You never hand-edit the wrong file again.</p><p>Want to include the root project in a workspace run? Use --include-workspace-root:</p><pre>npm run lint -ws --include-workspace-root --if-present</pre><p>And the escape hatch: --workspaces=false tells npm to ignore workspaces for a command, which is handy when a config file sets them on by default.</p><p>For smarter filtering, use npm query. It takes CSS-like selectors:</p><pre>npm query &quot;.workspace&quot;<br>npm query &quot;.workspace[name^=@acme]&quot;</pre><p>Both print a JSON array of matching packages. It’s a good way to feed names into a script, though I’d keep it for CI tooling and not daily use.</p><p><strong>Under the hood.</strong> -w doesn&#39;t change which tree gets built. The whole repo still has one dependency tree. The flag only narrows which packages the command runs against. That&#39;s also why installing with -w can still touch the shared lockfile and hoisted node_modules.</p><p>Here’s a quick mid-article check: do you target workspaces by name or by folder in your own scripts? Tell me which and why in the comments. I’m curious whether folder paths are still winning out there.</p><p>Filtering gets one package. What about running several at the same time?</p><h3>Can NPM workspaces run scripts in parallel?</h3><p>Not by themselves. --workspaces goes one at a time. Anyone claiming otherwise is describing a different tool.</p><h3>Quick version</h3><p>For long-running tasks like watch mode, add concurrently at the root:</p><pre>npm install -D concurrently</pre><p>Give each package a watch script:</p><pre>// acme/packages/core/package.json (scripts section)<br>{<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;tsc&quot;,<br>    &quot;dev&quot;: &quot;tsc --watch --preserveWatchOutput&quot;<br>  }<br>}</pre><pre>// acme/packages/app/package.json (scripts section)<br>{<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;tsc&quot;,<br>    &quot;dev&quot;: &quot;tsc --watch --preserveWatchOutput&quot;,<br>    &quot;start&quot;: &quot;node dist/index.js&quot;<br>  }<br>}</pre><p>Then wire both together from the root:</p><pre>// acme/package.json (scripts section)<br>{<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;npm run build -w @acme/core &amp;&amp; npm run build -w @acme/app&quot;,<br>    &quot;dev&quot;: &quot;concurrently -n core,app \&quot;npm run dev -w @acme/core\&quot; \&quot;npm run dev -w @acme/app\&quot;&quot;<br>  }<br>}</pre><p>Run npm run dev and you&#39;ll see interleaved, labeled output like:</p><pre>[core] Starting compilation in watch mode...<br>[app] Starting compilation in watch mode...<br>[core] Found 0 errors. Watching for file changes.<br>[app] Found 0 errors. Watching for file changes.</pre><h3>Real-world version</h3><p>Parallel is for <strong>independent</strong> work, like watchers and linters. Do not run builds in parallel when one package consumes another’s output. That’s the build-order bug again, but now it’s flaky instead of consistent, which is worse. Flaky bugs eat evenings.</p><p>My rule: sequential &amp;&amp; for builds, parallel for watchers.</p><p><strong>Under the hood.</strong> concurrently just spawns child processes and prefixes their output. Each npm run ... -w call is its own npm process doing its own tree lookup. That&#39;s fine for a handful of packages. If you find yourself orchestrating dozens with caching and a task graph, that&#39;s the point to look at a dedicated build tool, such as Turborepo or Nx, and keep npm workspaces underneath for installs. Workspaces handle installs and linking. They aren&#39;t a task runner, and that&#39;s okay.</p><p>Next up is the trap I teased earlier, the sibling of the stale dist folder.</p><h3>What’s the hoisting trap that hides missing dependencies?</h3><p>Because npm hoists shared dependencies to the root node_modules, a package can import something it never declared and still work.</p><h3>Quick version: the wrong version that still runs</h3><p>Say core uses dayjs and declares it. App imports it too, but forgot to add it to its own package.json:</p><pre>// acme/packages/app/src/index.ts<br>import dayjs from &quot;dayjs&quot;;<br>import { formatPrice } from &quot;@acme/core&quot;;</pre><pre>console.log(`${dayjs(&quot;2026-01-15&quot;).format(&quot;YYYY-MM-DD&quot;)}: ${formatPrice(1999)}`);</pre><pre>npm install dayjs -w @acme/core<br>npm run build<br>npm run start -w @acme/app</pre><p>Expected output:</p><pre>2026-01-15: $19.99</pre><p>It works, because dayjs sits in the root node_modules where app&#39;s Node resolution finds it. The bug is invisible until someone publishes @acme/app on its own, or moves it to another repo, and it explodes with &quot;module not found.&quot;</p><h3>Real-world version</h3><p>The fix is boring and correct: declare what you import.</p><pre>npm install dayjs -w @acme/app</pre><p>Now both packages list dayjs themselves. npm still stores one copy at the root if the versions agree, so you keep the disk savings and the honesty.</p><p>My opinion: this is the number one thing most developers get wrong with workspaces. They treat the root node_modules like a shared bucket. It&#39;s a cache of the tree, not a permission slip.</p><p><strong>Under the hood.</strong> Hoisting is a dedupe optimization in the tree builder, and Node’s resolution walks up parent folders looking for node_modules. The two together create &quot;phantom dependencies.&quot; Stricter package managers fight this by design. With npm you have to enforce the discipline yourself, for example through a lint rule or a quick review habit.</p><p>Last stretch. Workspaces have more commands that keep your packages consistent with each other.</p><h3>What other workspace features keep packages in sync?</h3><p>Most npm commands accept -w or -ws. Four are worth knowing.</p><p><strong>Scaffold a new package.</strong> npm init -w creates the folder if it doesn&#39;t exist and adds it to the root workspaces setup:</p><pre>npm init -w ./packages/utils -y</pre><p><strong>Edit </strong><strong>package.json fields everywhere.</strong> This is my favorite consistency tool:</p><pre>npm pkg set engines.node=&quot;&gt;=20&quot; --workspaces<br>npm pkg get version --workspaces</pre><p>The first sets the same Node engine range in every package. The second prints each package’s version, keyed by name.</p><p><strong>Version and publish selectively:</strong></p><pre>npm version patch -w @acme/core<br>npm publish -w @acme/core</pre><p><strong>Run any binary in one package:</strong></p><pre>npm exec -w @acme/app -- tsc --noEmit</pre><p>Here’s the full running example one more time, the root file you should end up with:</p><pre>// acme/package.json<br>{<br>  &quot;name&quot;: &quot;acme&quot;,<br>  &quot;version&quot;: &quot;1.0.0&quot;,<br>  &quot;private&quot;: true,<br>  &quot;workspaces&quot;: [&quot;packages/*&quot;],<br>  &quot;scripts&quot;: {<br>    &quot;build&quot;: &quot;npm run build -w @acme/core &amp;&amp; npm run build -w @acme/app&quot;,<br>    &quot;dev&quot;: &quot;concurrently -n core,app \&quot;npm run dev -w @acme/core\&quot; \&quot;npm run dev -w @acme/app\&quot;&quot;,<br>    &quot;typecheck&quot;: &quot;npm run build --workspaces --if-present -- --noEmit&quot;<br>  },<br>  &quot;devDependencies&quot;: {<br>    &quot;@types/node&quot;: &quot;^22.0.0&quot;,<br>    &quot;concurrently&quot;: &quot;^9.0.0&quot;,<br>    &quot;typescript&quot;: &quot;^5.6.0&quot;<br>  }<br>}</pre><p>I kept typecheck out of the earlier sections on purpose. Try it, and think about why --noEmit against app would fail on a clean checkout (hint: same cause as before, no core types without a core build). That&#39;s your homework.</p><h3>Bonus tips, one per level</h3><ul><li><strong>Beginner:</strong> run npm ls --workspaces from the root to see every package and its dependencies at a glance.</li><li><strong>Real-world:</strong> commit the root package-lock.json and use npm ci in CI. One lockfile means one source of truth.</li><li><strong>Under the hood:</strong> if a package behaves differently locally than in CI, delete node_modules and every dist, run npm ci, and rebuild. That recreates the clean-checkout world on your own machine.</li></ul><h3>Back to that red pipeline</h3><p>Remember 6:40 pm and the missing @acme/core? Your laptop had a stale dist folder quietly doing the job that build ordering should have done. CI had no such luxury. Workspaces gave you linking, one lockfile, and one install. They did not give you a dependency-aware build order, and they never promised to.</p><p>Here’s the recap:</p><ol><li>Workspaces link local packages through one root install and one lockfile.</li><li>-workspaces runs scripts one by one, in npm&#39;s order, not your dependency order.</li><li>w filters by name, path, or parent folder, and works on install, version, publish, and exec.</li><li>Parallel work needs a helper like concurrently, and it belongs to watchers, not builds.</li><li>Hoisting hides undeclared imports. Declare what you import.</li><li>npm pkg set --workspaces keeps shared metadata identical everywhere.</li></ol><p><strong>What did you think?</strong></p><p>Did this approach match how you are solving it, or do you have a different take? Drop a comment. I genuinely read every single one.</p><p><strong>Found this helpful?</strong></p><p>If this saved you even a few minutes of debugging or confusion, hit that clap button so others can find it too. It really does make a difference.</p><p><strong>Want more tips like this?</strong></p><p>I share one practical dev insight every week. Follow me here on Medium or subscribe to my newsletter so you never miss one.</p><p>Let us connect. Find me on LinkedIn or GitHub and let us keep the conversation going.</p><p>Follow Me for More Angular &amp; Frontend Goodness:</p><p>I regularly share hands-on tutorials, clean code tips, scalable frontend architecture, and real-world problem-solving guides.</p><ul><li>💼 <a href="https://www.linkedin.com/in/errajatmalik/"><strong>LinkedIn</strong></a> — Let’s connect professionally</li><li>🎥 <a href="https://www.threads.net/@er.rajatmalik"><strong>Threads</strong></a> — Short-form frontend insights</li><li>🐦 <a href="https://twitter.com/er_rajatmalik"><strong>X (Twitter)</strong></a> — Developer banter + code snippets</li><li>👥 <a href="http://devrajat.bsky.social/"><strong>BlueSky</strong></a> — Stay up to date on frontend trends</li><li>🌟 <a href="https://github.com/malikrajat"><strong>GitHub Projects</strong></a> — Explore code in action</li><li>🌐 <a href="https://malikrajat.github.io/"><strong>Website</strong></a> — Everything in one place</li><li>📚 <a href="https://medium.com/@codewithrajat/"><strong>Medium Blog</strong></a> — Long-form content and deep-dives</li><li>💬 <a href="https://dev.to/codewithrajat"><strong>Dev Blog</strong></a> — Free Long-form content and deep-dives</li><li>✉️ <a href="https://codewithrajat.substack.com/"><strong>Substack</strong></a> — Weekly frontend stories &amp; curated resources</li><li>🧩 <a href="https://rajatmalik.dev/"><strong>Portfolio</strong></a> — Projects, talks, and recognitions</li><li>✍️ <a href="https://hashnode.com/@codeswithrajat"><strong>Hashnode</strong></a> — Developer blog posts &amp; tech discussions</li><li>✍️ <a href="https://www.reddit.com/user/codewithrajat/submitted/">Reddit</a> — Developer blog posts &amp; tech discussions</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49bd8370b3c1" width="1" height="1" alt=""><hr><p><a href="https://javascript.plainenglish.io/npm-workspaces-why-does-my-monorepo-build-work-locally-but-break-in-ci-49bd8370b3c1">NPM Workspaces: Why Does My Monorepo Build Work Locally but Break in CI?</a> was originally published in <a href="https://javascript.plainenglish.io">JavaScript in Plain English</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Future of Frontend Development Might Have Less JavaScript in It]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/the-future-of-frontend-development-might-have-less-javascript-in-it-32c009c9aa90?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1672/1*dVhHkE3h68sI4xX5yE5bRw.png" width="1672"></a></p><p class="medium-feed-snippet">The web is evolving. Server Components, native browser APIs, HTML improvements, and server-side rendering are challenging the idea that&#x2026;</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/the-future-of-frontend-development-might-have-less-javascript-in-it-32c009c9aa90?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/the-future-of-frontend-development-might-have-less-javascript-in-it-32c009c9aa90?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/32c009c9aa90</guid>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[javascript-tips]]></category>
            <category><![CDATA[javascript-development]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[javascript]]></category>
            <dc:creator><![CDATA[Huzair Awan]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 15:33:26 GMT</pubDate>
            <atom:updated>2026-11-17T15:33:25.553Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Node 27 Caps Your Heap at Half Your RAM by Default. Measure Before You Ship.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/node-27-caps-your-heap-at-half-your-ram-by-default-measure-before-you-ship-990972d6dd44?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1600/1*pwsFzKmadg_414byfg80ng.png" width="1600"></a></p><p class="medium-feed-snippet">Upgrade a service to Node 27 and, on a typical 8 GB box, your heap can grow to 4 GB before V8 collects hard. On Node 26 that ceiling was 2&#x2026;</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/node-27-caps-your-heap-at-half-your-ram-by-default-measure-before-you-ship-990972d6dd44?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/node-27-caps-your-heap-at-half-your-ram-by-default-measure-before-you-ship-990972d6dd44?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/990972d6dd44</guid>
            <category><![CDATA[v8]]></category>
            <category><![CDATA[backend-development]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[nodejs]]></category>
            <dc:creator><![CDATA[Sohail Saifi]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 14:52:41 GMT</pubDate>
            <atom:updated>2026-11-17T14:52:40.275Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[I Got Another Job. I’m Not Calling It Security.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/i-got-another-job-im-not-calling-it-security-8b4589094f82?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1802/1*iBk_3KkKIC8N6lrL7HKHGw.png" width="1802"></a></p><p class="medium-feed-snippet">After almost a year of unemployment, I&#x2019;m grateful for a salary again. I&#x2019;m also asking it to do less.</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/i-got-another-job-im-not-calling-it-security-8b4589094f82?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/i-got-another-job-im-not-calling-it-security-8b4589094f82?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/8b4589094f82</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Yusuf Seyitoğlu]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 11:23:29 GMT</pubDate>
            <atom:updated>2026-11-17T11:23:28.263Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[I Installed 8 Everyday npm Packages. 384 Arrived. Node Has a Built-in for Each.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/i-installed-8-everyday-npm-packages-384-arrived-node-has-a-built-in-for-each-6b1871798e1f?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1536/1*o8N_Wq8nKUE-T8oQpze7vw.png" width="1536"></a></p><p class="medium-feed-snippet">Tests, watch mode, .env files, CLI flags, colours, globs, HTTP and TypeScript, redone with zero installs. Plus where the package still&#x2026;</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/i-installed-8-everyday-npm-packages-384-arrived-node-has-a-built-in-for-each-6b1871798e1f?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/i-installed-8-everyday-npm-packages-384-arrived-node-has-a-built-in-for-each-6b1871798e1f?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/6b1871798e1f</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[nodejs]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[web-development]]></category>
            <dc:creator><![CDATA[Ferid Brković]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 10:15:32 GMT</pubDate>
            <atom:updated>2026-11-17T10:15:31.100Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[JavaScript Memory Leaks You Might Be Creating Without Knowing]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/javascript-memory-leaks-you-might-be-creating-without-knowing-90b394b954d5?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1408/1*51kP0v2evOGF--hzZRmzgQ.jpeg" width="1408"></a></p><p class="medium-feed-snippet">At the beginning, your JavaScript app can be working great.
Pages load fast. Buttons react immediately. API calls are successful.
However&#x2026;</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/javascript-memory-leaks-you-might-be-creating-without-knowing-90b394b954d5?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/javascript-memory-leaks-you-might-be-creating-without-knowing-90b394b954d5?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/90b394b954d5</guid>
            <category><![CDATA[javascript-tips]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[javascript-frameworks]]></category>
            <category><![CDATA[javascript-development]]></category>
            <dc:creator><![CDATA[Dr Abdullah Azhar]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 10:09:36 GMT</pubDate>
            <atom:updated>2026-11-17T10:09:34.867Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Code Is Cheap. Trust Is Expensive.]]></title>
            <link>https://javascript.plainenglish.io/code-is-cheap-trust-is-expensive-635238ef2b10?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/635238ef2b10</guid>
            <category><![CDATA[code-review]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[developer-productivity]]></category>
            <dc:creator><![CDATA[Alamgir Hossen]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 10:07:43 GMT</pubDate>
            <atom:updated>2026-11-17T10:07:42.505Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>Why Verification Is Becoming the New Bottleneck in AI Software Development</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vu1qK4F1gRTyQo_4UGgIAQ.png" /><figcaption>A software developer reviewing AI-generated code to verify its quality, security, and reliability.</figcaption></figure><p>Imagine asking an AI coding assistant to build a feature.</p><p>Within seconds, it generates several hundred lines of code. It creates the API endpoint, writes the database queries, handles validation, and even generates a few tests.</p><p>The implementation looks clean. The structure makes sense. The code compiles.</p><p>A few years ago, completing the same task might have taken hours.</p><p>But there is one question we often forget to ask:</p><p><strong>How do we know the code is actually correct?</strong></p><p>Does it satisfy the requirements? Does it follow the existing architecture? Can it handle unexpected inputs? Is it secure? Will it break something that was already working?</p><p>Generating code is becoming easier. Trusting that code is an entirely different challenge.</p><p>AI has dramatically changed how developers write software, but it has not eliminated the responsibility of building reliable software. In many ways, it has shifted where engineering effort is required.</p><p>And that shift is creating a new bottleneck in software development: verification.</p><h3>Code Generation Is No Longer the Hardest Part</h3><p>For decades, writing code was one of the most time-consuming parts of software development.</p><p>Developers had to understand requirements, design solutions, write implementations, debug errors, and maintain existing systems.</p><p>AI coding assistants have changed that workflow.</p><p>Today, developers can describe a feature in natural language and receive an implementation almost instantly. They can generate API endpoints, database models, frontend components, unit tests, and documentation with significantly less manual effort.</p><p>This is undoubtedly valuable.</p><p>However, software engineering has never been simply about producing lines of code.</p><p>Consider a typical full-stack feature. An API endpoint might need to:</p><ul><li>Validate incoming data.</li><li>Check user authentication and authorization.</li><li>Interact with the correct database records.</li><li>Handle database transactions.</li><li>Respect existing business rules.</li><li>Return predictable error responses.</li><li>Maintain compatibility with existing consumers.</li></ul><p>An AI model can generate an implementation for all of these requirements.</p><p>But generating an implementation does not prove that every requirement has been satisfied.</p><p>The code might work for the example provided in the prompt while failing under conditions that were never mentioned.</p><p>It might follow the requested implementation while violating an architectural decision elsewhere in the project.</p><p>It might even solve the immediate problem while introducing a more complicated maintenance problem.</p><p><strong>The difference between generating code and engineering software is the ability to establish that the solution actually works.</strong></p><h3>The Growing Gap Between AI Output and Developer Trust</h3><p>Interestingly, developers are not blindly trusting AI-generated output.</p><p>The Stack Overflow Developer Survey 2026 provides some useful evidence.</p><p>Among respondents answering the AI trust question:</p><ul><li>48% said they trust AI when they can easily verify its output.</li><li>16.3% trust it for many tasks, but not important work decisions.</li><li>16.3% trust it for low-risk work tasks only.</li><li>Only 6.6% said they trust AI for many tasks, including important work decisions.</li></ul><p>These findings reveal something important about modern software development.</p><p>Developers are willing to use AI, but their trust is conditional.</p><p>They want to understand what the AI has produced before allowing that output to influence important decisions.</p><p>The same survey also found that developers actively verify AI-generated information rather than accepting it without examination.</p><p>For example, 76.5% reported running AI-generated output locally, while 63.9% compared it with existing code or repositories.</p><p>This suggests that AI-assisted development does not eliminate developer involvement. It changes the nature of that involvement.</p><p>Developers are spending less time writing some implementations from scratch and more time evaluating whether generated implementations deserve to be accepted.</p><p><a href="https://survey.stackoverflow.co/2026">Source: Stack Overflow Developer Survey 2026</a></p><h3>AI Makes Developers Faster, but Where Does the Saved Time Go?</h3><p>The productivity conversation becomes more interesting when we look beyond code generation.</p><p>Atlassian’s State of Developer Experience Report 2026 surveyed more than 1,500 software practitioners.</p><p>The findings present an interesting contrast:</p><ul><li>87% of developers say AI helps them ship work faster.</li><li>72% report experiencing the same or more friction in their daily workflow.</li><li>49% spend four or more hours per week reviewing AI output, correcting prompts, and performing related work.</li><li>75% say they have greater responsibility for validating the quality and safety of AI-generated output.</li></ul><p>These numbers suggest that faster implementation does not automatically translate into a frictionless development process.</p><p>Developers may complete individual coding tasks faster, but the overall development workflow still involves understanding requirements, reviewing changes, fixing problems, and coordinating with other people.</p><p>Consider what happens when an AI agent generates a feature across ten files.</p><p>The developer still needs to understand:</p><ul><li>Why each file was modified.</li><li>Whether the changes are necessary.</li><li>Whether the implementation introduces duplication.</li><li>Whether existing functionality remains intact.</li><li>Whether the tests cover the actual requirements.</li></ul><p>The implementation may take seconds to generate, but understanding its impact can take considerably longer.</p><p>This is an important distinction.</p><p><strong>AI can accelerate individual tasks without necessarily accelerating the entire software delivery process.</strong></p><p><a href="https://www.atlassian.com/blog/development/developer-experience-report-2026">Source: Atlassian State of Developer Experience Report 2026</a></p><h3>The Hidden Risks Inside AI-Generated Code</h3><p>One of the biggest mistakes developers can make is assuming that readable code is reliable code.</p><p>AI-generated implementations often look convincing. They can follow familiar patterns, use appropriate naming conventions, and produce clean-looking structures.</p><p>But correctness is not always visible from appearance.</p><p>There are several areas where problems can remain hidden.</p><h3>1. Plausible but Incorrect Logic</h3><p>AI can produce code that looks reasonable but misunderstands the actual business requirement.</p><p>Imagine a payment-processing endpoint.</p><p>The AI generates an implementation that validates the payment amount and updates the transaction status.</p><p>Everything appears to work during basic testing.</p><p>But what happens when the same payment request is submitted twice?</p><p>What happens when the database update succeeds but the external payment provider times out?</p><p>What happens when two requests attempt to update the same transaction simultaneously?</p><p>These situations require more than generating syntactically correct code. They require understanding system behavior, concurrency, idempotency, and failure scenarios.</p><h3>2. Missing Edge Cases</h3><p>AI-generated code often performs well when the input closely matches the examples in the prompt.</p><p>Real applications rarely behave that predictably.</p><p>Users submit unexpected values. External services fail. Database records disappear. Requests arrive in an unexpected order.</p><p>A function that works perfectly with valid input might fail when it receives null values, duplicated records, incomplete data, or unexpected states.</p><p>These cases are not always obvious from the implementation.</p><h3>3. Architectural Inconsistency</h3><p>An application is not just a collection of independent functions.</p><p>It is a system with established conventions, dependencies, shared services, and architectural decisions.</p><p>An AI agent might generate a perfectly reasonable implementation that does not belong in the existing architecture.</p><p>For example, it might introduce a new utility when the application already has a shared service for that responsibility.</p><p>It might duplicate validation logic, create unnecessary abstractions, or bypass an existing permission-checking mechanism.</p><p>The code may work today while making the application harder to maintain tomorrow.</p><h3>4. Security Vulnerabilities</h3><p>Security is particularly important because a vulnerable implementation can look completely functional.</p><p>A 2025 empirical study published in ACM’s <em>Transactions on Software Engineering and Methodology</em> examined 733 AI-generated code snippets collected from GitHub projects.</p><p>The researchers identified security weaknesses in 27.3% of the analyzed snippets.</p><p>The findings included:</p><ul><li>29.5% of the analyzed Python snippets contained security weaknesses.</li><li>24.2% of the analyzed JavaScript snippets contained security weaknesses.</li><li>The researchers identified 628 weaknesses across 43 CWE categories.</li></ul><p>These results should not be interpreted as saying that 27.3% of all modern AI-generated code is insecure. The study examined a particular dataset and specific code-generation tools.</p><p>However, it illustrates an important engineering principle: code that appears functional can still contain security weaknesses.</p><p>For full-stack developers, this means paying particular attention to authentication, authorization, input validation, SQL injection, XSS, sensitive data handling, and dependency security.</p><p><a href="https://doi.org/10.1145/3716848">Source: Security Weaknesses of Copilot-Generated Code in GitHub Projects: An Empirical Study</a></p><h3>More Code Does Not Always Mean More Productivity</h3><p>There is another important question worth discussing.</p><p>If AI can generate code faster, does that necessarily make developers more productive?</p><p>Not always.</p><p>In 2025, METR published a randomized controlled trial examining the impact of early-2025 AI coding tools on experienced open-source developers.</p><p>The experiment involved 16 developers completing 246 tasks in mature repositories they already knew well.</p><p>Before the experiment, developers expected AI tools to make them considerably faster.</p><p>However, the observed results were different.</p><p>Allowing AI tools actually increased task completion time by 19% in that particular experiment.</p><p>This does not mean AI universally makes developers slower. The research was relatively small, focused on experienced developers working in familiar, complex repositories, and tested tools from an earlier generation.</p><p>Nevertheless, it challenges the assumption that introducing AI automatically increases productivity.</p><p>Why might this happen?</p><p>Because using AI introduces additional work:</p><ol><li>Writing and refining prompts.</li><li>Providing enough context.</li><li>Reviewing generated implementations.</li><li>Correcting incorrect assumptions.</li><li>Running and debugging the changes.</li><li>Integrating the result into an existing codebase.</li></ol><p>Sometimes the AI saves more time than it consumes.</p><p>Sometimes it does not.</p><p>The important thing is to evaluate the complete development cycle rather than measuring how quickly code appears on the screen.</p><p><a href="https://metr.org/Early_2025_AI_Experienced_OS_Devs_Study-paper.pdf">Source: METR — Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity</a></p><h3>Verification Is Becoming a Core Engineering Skill</h3><p>If generating code is becoming cheaper, developers need to become better at evaluating what gets generated.</p><p>This does not mean manually inspecting every character of every AI-generated file.</p><p>It means building a reliable verification process around AI-assisted development.</p><p>Here is a practical workflow that I believe every developer working with AI coding agents should consider.</p><h3>1. Define the Expected Behavior Before Generating Code</h3><p>One of the most effective ways to reduce incorrect AI output is to make the requirements explicit.</p><p>Before asking AI to implement a feature, explain:</p><ul><li>What the feature should do.</li><li>Which existing modules it should interact with.</li><li>What business rules must be followed.</li><li>Which conditions should produce errors.</li><li>What should happen in unexpected situations.</li></ul><p>The clearer the expected behavior, the easier it becomes to verify the implementation.</p><p>A vague prompt produces more opportunities for incorrect assumptions.</p><h3>2. Review the Implementation Plan</h3><p>Do not always allow an AI agent to jump directly into implementation.</p><p>For larger changes, ask it to explain the proposed approach first.</p><p>Review which files it plans to modify, which dependencies it intends to introduce, how it will interact with existing services, and whether the approach respects the application’s architecture.</p><p>This gives developers an opportunity to identify problems before they become part of the codebase.</p><h3>3. Inspect the Actual Changes</h3><p>Never rely entirely on an AI agent’s summary of what it has done.</p><p>Inspect the actual Git diff.</p><p>Look for unnecessary changes, duplicated logic, unexpected dependencies, missing validation, and modifications to unrelated functionality.</p><p>The difference matters.</p><p>An agent might report that it implemented a feature successfully, but the actual changes may reveal that it modified several unrelated modules.</p><p>Your responsibility is to review what changed, not simply what the agent claims it changed.</p><h3>4. Run Meaningful Automated Tests</h3><p>Testing is one of the most important verification mechanisms.</p><p>Depending on the application, this should include:</p><ul><li>Unit tests for individual functions.</li><li>Integration tests for interacting components.</li><li>End-to-end tests for important user workflows.</li><li>Negative tests for invalid inputs.</li><li>Edge-case tests for unexpected conditions.</li></ul><p>AI can also help generate test cases.</p><p>However, there is an important limitation.</p><p>If the same AI that implemented the feature also creates tests based on the same incorrect assumptions, those tests may simply confirm the mistake.</p><p>Tests should be derived from independently understood requirements and expected behavior.</p><p>Passing tests provide evidence of correctness within the tested scenarios. They do not prove that the implementation is universally correct.</p><h3>5. Perform Security and Quality Checks</h3><p>Automated tools can identify problems that are easy to miss during manual review.</p><p>A good verification pipeline should include:</p><ul><li>Type checking.</li><li>Linting.</li><li>Static code analysis.</li><li>Dependency vulnerability scanning.</li><li>Secret detection.</li><li>Appropriate application security tests.</li></ul><p>For example, a TypeScript application should run its type checker rather than relying only on the editor’s apparent absence of errors.</p><p>A backend application should verify authorization rules and validate untrusted input.</p><p>A database operation involving multiple writes should receive appropriate transaction testing.</p><p>Automation does not replace engineering judgment, but it provides another layer of protection.</p><h3>6. Verify Again Before Production</h3><p>A successful local implementation is not necessarily production-ready.</p><p>Before deployment, verify the changes in an environment that resembles production as closely as practical.</p><p>Check important user workflows, database behavior, API compatibility, environment configuration, and failure handling.</p><p>Also consider monitoring and rollback strategies.</p><p>Ultimately, developers and engineering teams remain responsible for what they ship, regardless of who or what generated the code.</p><h3>Even Large-Scale Research Shows Why Context Matters</h3><p>Interestingly, recent research involving Meta’s software development environment offers another perspective.</p><p>A preliminary analysis presented at ESEM 2026 examined more than 10,000 developers and 400,000 code diffs.</p><p>The researchers found that AI-assisted code was associated with shorter review times and fewer production incidents.</p><p>At first glance, that sounds like clear evidence that AI-generated code is better.</p><p>But the researchers identified an important factor.</p><p>AI-assisted code was more likely to appear in less central parts of the codebase, larger changes, and test files. Once these contextual differences were considered, the apparent advantages became more complicated.</p><p>This matters because not every code change carries the same level of risk.</p><p>Changing a small UI component is fundamentally different from modifying a payment-processing system, authentication service, or core database logic.</p><p>A comparison that ignores these differences can produce misleading conclusions.</p><p>The lesson is not that AI-generated code is good or bad.</p><p>The lesson is that software quality must be evaluated in context.</p><p><a href="https://doi.org/10.4230/LIPIcs.ESEM.2026.93">Source: A Preliminary Analysis of the Impact of AI Assisted/Generated Code on Quality, Centrality and Review Time at Scale — ESEM 2026</a></p><h3>The Developer’s Role Is Changing</h3><p>For a long time, developers were often evaluated by their ability to implement features, solve technical problems, and deliver code efficiently.</p><p>Those skills still matter.</p><p>But AI is changing the balance of responsibilities.</p><p>When implementation becomes easier, deeper engineering judgment becomes more valuable.</p><p>Developers need to become better at:</p><ul><li>Understanding complex existing systems.</li><li>Evaluating unfamiliar code.</li><li>Identifying incorrect assumptions.</li><li>Designing meaningful tests.</li><li>Recognizing security risks.</li><li>Debugging complex failures.</li><li>Making architectural decisions.</li><li>Maintaining software over time.</li></ul><p>The most effective developers will not necessarily be the ones who generate the most code.</p><p>They will be the ones who understand what should be built, know how to use AI effectively, and can establish whether the resulting implementation is reliable.</p><p>This also changes how teams should evaluate productivity.</p><p>Counting generated lines of code, accepted AI suggestions, or the number of prompts used does not tell us whether a software team is building better products.</p><p>A better measurement focuses on outcomes: successful delivery, defect rates, reliability, maintainability, and the ability to respond to changing requirements.</p><h3>The Future Belongs to Developers Who Can Verify</h3><p>AI coding tools are not going away.</p><p>They will continue improving, handling more complicated tasks, understanding larger codebases, and participating in increasingly sophisticated development workflows.</p><p>That is a positive development.</p><p>But there is a difference between producing software and producing software that deserves to be trusted.</p><p>The faster we generate code, the more important it becomes to understand what that code actually does.</p><p>Verification should not be treated as an additional step that developers perform only when they have extra time. It is part of software engineering itself.</p><p>AI can write the implementation.</p><p>AI can suggest the architecture.</p><p>AI can generate the tests.</p><p>AI can even help investigate potential security problems.</p><p>But accepting the result without understanding its implications is not engineering.</p><p>It is simply hoping that generated code is correct.</p><p><strong>The future of software engineering won’t belong to developers who generate the most code. It will belong to developers who know how to verify it, challenge it, and take responsibility for the code they ship.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=635238ef2b10" width="1" height="1" alt=""><hr><p><a href="https://javascript.plainenglish.io/code-is-cheap-trust-is-expensive-635238ef2b10">Code Is Cheap. Trust Is Expensive.</a> was originally published in <a href="https://javascript.plainenglish.io">JavaScript in Plain English</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Layoff Emergency Kit: 7 Things I Keep Ready While I Still Have a Job]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://javascript.plainenglish.io/my-layoff-emergency-kit-7-things-i-keep-ready-while-i-still-have-a-job-da26c6f2107d?source=rss----4b3a1ed4f11c---4"><img src="https://cdn-images-1.medium.com/max/1600/1*aI18bL9e2RzsKXnOI6iFCQ.png" width="1600"></a></p><p class="medium-feed-snippet">Nobody gets a warning. So I stopped waiting for one.</p><p class="medium-feed-link"><a href="https://javascript.plainenglish.io/my-layoff-emergency-kit-7-things-i-keep-ready-while-i-still-have-a-job-da26c6f2107d?source=rss----4b3a1ed4f11c---4">Continue reading on JavaScript in Plain English »</a></p></div>]]></description>
            <link>https://javascript.plainenglish.io/my-layoff-emergency-kit-7-things-i-keep-ready-while-i-still-have-a-job-da26c6f2107d?source=rss----4b3a1ed4f11c---4</link>
            <guid isPermaLink="false">https://medium.com/p/da26c6f2107d</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[personal-finance]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[layoffs]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[The woman]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 10:01:15 GMT</pubDate>
            <atom:updated>2026-11-17T10:01:13.951Z</atom:updated>
        </item>
    </channel>
</rss>