<?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: Alex Mantello</title>
    <description>The latest articles on DEV Community by Alex Mantello (@jamaalx).</description>
    <link>https://dev.to/jamaalx</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%2F3833071%2Fa59fd45a-8349-458c-9588-2e2ef18e5933.png</url>
      <title>DEV Community: Alex Mantello</title>
      <link>https://dev.to/jamaalx</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/jamaalx"/>
    <language>en</language>
    <item>
      <title>AI slop isn't a model problem. It's what you get when nothing checks the output before it ships.</title>
      <dc:creator>Alex Mantello</dc:creator>
      <pubDate>Tue, 17 Nov 2026 00:08:08 +0000</pubDate>
      <link>https://dev.to/jamaalx/ai-slop-isnt-a-model-problem-its-what-you-get-when-nothing-checks-the-output-before-it-ships-2hnp</link>
      <guid>https://dev.to/jamaalx/ai-slop-isnt-a-model-problem-its-what-you-get-when-nothing-checks-the-output-before-it-ships-2hnp</guid>
      <description>&lt;p&gt;Claude is good enough now that most slop isn't the model's fault. It's the workflow: one prompt, no facts, no checks, straight to publish.&lt;/p&gt;

&lt;p&gt;I've spent the last few months building the boring part that comes after generation, around Claude Code. These are the five things that actually cut slop for us.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. Facts live in a file, not in the prompt.&lt;/strong&gt;&lt;br&gt;
Every workflow reads one facts file: what we sell, prices, what we don't do. The rule in the system prompt is one line: if it isn't in the file, say you don't know. Most "the AI invented a feature" problems disappeared that day.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. A linter for prose.&lt;/strong&gt;&lt;br&gt;
Before a human sees anything, a deterministic checker rejects the draft and tells the agent why:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;emoji, unfilled placeholders like [CLIENT NAME]&lt;/li&gt;
&lt;li&gt;"delve", "seamless", "game-changer", "unlock" and friends&lt;/li&gt;
&lt;li&gt;wrong length for the platform, wrong language, missing link
The agent fixes it and resubmits. No model call needed for any of this, and it catches more than you'd expect.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;3. A second look by a model, then a person.&lt;/strong&gt;&lt;br&gt;
What passes the linter gets a quick model review against the brand rules, then lands in an approval queue. Nothing leaves (post, email, client file) until someone clicks Approve. Rejections need a reason, and the agent sees it next time.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4. Evals per workflow.&lt;/strong&gt;&lt;br&gt;
Every workflow ships with test cases: an input, plus what a good output must and must not contain. A model grades them before a workflow goes live. Our internal set is 345 cases. When we change a prompt, we rerun them instead of eyeballing two examples.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;5. Small toolboxes.&lt;/strong&gt;&lt;br&gt;
Agents that run unattended get an allowlist, not a denylist:&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;"allow": ["Bash(ourcli:*)", "Read(./work/**)", "Edit(./work/**)"],
"deny":  ["Bash(sudo:*)", "Bash(curl:*)", "Bash(npm install:*)", "Bash(pip install:*)"]
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

&lt;p&gt;with &lt;code&gt;--permission-mode dontAsk&lt;/code&gt;. Less clever, much easier to trust.&lt;/p&gt;

&lt;p&gt;None of this is fancy. It's the same thing you'd do for a junior hire: give them the facts, a style guide, a reviewer and limited access.&lt;/p&gt;

&lt;p&gt;Disclosure: this is how Zen Staff works, the app I build. It started as my fork of T3 Code in July. It runs the official Claude Code (or Codex) on your own account, it's free and open source (MIT). Windows and Linux for now, no Mac app yet.&lt;/p&gt;

&lt;p&gt;Curious what others do. What's the one check that removed the most slop from your agents' output?&lt;/p&gt;

</description>
      <category>ai</category>
      <category>claude</category>
      <category>productivity</category>
      <category>opensource</category>
    </item>
    <item>
      <title>I built a Claude Code plugin that auto-names your conversations</title>
      <dc:creator>Alex Mantello</dc:creator>
      <pubDate>Thu, 19 Mar 2026 04:06:59 +0000</pubDate>
      <link>https://dev.to/jamaalx/i-built-a-claude-code-plugin-that-auto-names-your-conversations-1lcl</link>
      <guid>https://dev.to/jamaalx/i-built-a-claude-code-plugin-that-auto-names-your-conversations-1lcl</guid>
      <description>&lt;p&gt;Claude Code saves all conversations as random UUID filenames like &lt;code&gt;c348d36f-8002-40ca-8523-17821d6a0bea.jsonl&lt;/code&gt;. After 100+ sessions,&lt;br&gt;
  finding an old conversation is impossible.&lt;/p&gt;

&lt;p&gt;I searched npm and the Claude Code plugin marketplace - nothing existed to solve this.&lt;/p&gt;

&lt;p&gt;## The solution&lt;/p&gt;

&lt;p&gt;I built &lt;strong&gt;chat-namer&lt;/strong&gt;, an open-source Claude Code plugin that:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Auto-names&lt;/strong&gt; every conversation from the first user message via a Stop hook&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Search&lt;/strong&gt; all your chats by keyword&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Resume&lt;/strong&gt; any old conversation without memorizing UUIDs&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Rename&lt;/strong&gt; conversations with custom names&lt;/li&gt;
&lt;li&gt;Works as a &lt;strong&gt;Claude Code plugin&lt;/strong&gt; with slash commands or as a &lt;strong&gt;standalone CLI&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;## How it works&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Scans &lt;code&gt;~/.claude/projects/&lt;/code&gt; for conversation &lt;code&gt;.jsonl&lt;/code&gt; files&lt;/li&gt;
&lt;li&gt;Reads the first user message to generate a readable name&lt;/li&gt;
&lt;li&gt;Stores names in a separate &lt;code&gt;chat-names.json&lt;/code&gt; file&lt;/li&gt;
&lt;li&gt;Never modifies conversation files - names are just metadata&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;## Install&lt;/p&gt;

&lt;p&gt;As a Claude Code plugin:&lt;/p&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;
bash
  git clone https://github.com/Jamaalx/claude-chat-namer.git
  claude --plugin-dir ./claude-chat-namer

  Or standalone via npm:
  npx claude-chat-namer

  Slash commands

  ┌────────────────────────────┬────────────────────────────────────────┐
  │          Command           │              Description               │
  ├────────────────────────────┼────────────────────────────────────────┤
  │ /chat-namer:list           │ List all conversations with names      │
  ├────────────────────────────┼────────────────────────────────────────┤
  │ /chat-namer:search &amp;lt;query&amp;gt; │ Search by keyword                      │
  ├────────────────────────────┼────────────────────────────────────────┤
  │ /chat-namer:resume &amp;lt;query&amp;gt; │ Find a chat and get the resume command │
  ├────────────────────────────┼────────────────────────────────────────┤
  │ /chat-namer:rename         │ Rename a conversation                  │
  ├────────────────────────────┼────────────────────────────────────────┤
  │ /chat-namer:auto-name      │ Auto-name all unnamed conversations    │
  └────────────────────────────┴────────────────────────────────────────┘

  Zero dependencies

  The entire plugin uses only Node.js built-ins: fs, path, os, readline. No npm install needed.

  Links

  - GitHub: https://github.com/Jamaalx/claude-chat-namer
  - npm: https://www.npmjs.com/package/claude-chat-namer

  First plugin of its kind. Published on npm and submitted to the official Anthropic plugin marketplace.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

</description>
      <category>claude</category>
      <category>opensource</category>
      <category>javascript</category>
      <category>ai</category>
    </item>
  </channel>
</rss>
