<?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>Forem: Saniel Doe</title>
    <description>The latest articles on Forem by Saniel Doe (@saniel_doe_7c469c36088a98).</description>
    <link>https://forem.com/saniel_doe_7c469c36088a98</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.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3883281%2Ffa14420c-9f5a-4de3-bd9b-12b0a9b88d3f.png</url>
      <title>Forem: Saniel Doe</title>
      <link>https://forem.com/saniel_doe_7c469c36088a98</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://forem.com/feed/saniel_doe_7c469c36088a98"/>
    <language>en</language>
    <item>
      <title>I built the most honest AI coding agent: it tells you exactly what it did (nothing)</title>
      <dc:creator>Saniel Doe</dc:creator>
      <pubDate>Thu, 16 Apr 2026 22:42:58 +0000</pubDate>
      <link>https://forem.com/saniel_doe_7c469c36088a98/i-built-the-most-honest-ai-coding-agent-it-tells-you-exactly-what-it-did-nothing-35ib</link>
      <guid>https://forem.com/saniel_doe_7c469c36088a98/i-built-the-most-honest-ai-coding-agent-it-tells-you-exactly-what-it-did-nothing-35ib</guid>
      <description>&lt;p&gt;I've been using AI coding agents for a while. They all share one thing:&lt;br&gt;
  they sound very confident about work they may or may not have done.                                                   &lt;/p&gt;

&lt;p&gt;So I built one that's upfront about it.&lt;/p&gt;
&lt;h2&gt;
  
  
  Meet p_ (Pinocchio)
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Asks smart clarifying questions (powered by a real LLM)
&lt;/li&gt;
&lt;li&gt;Generates a convincing implementation plan
&lt;/li&gt;
&lt;li&gt;Shows professional tool call output
&lt;/li&gt;
&lt;li&gt;Does absolutely nothing
&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The nose grows with every "completed" task.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fgithub.com%2Fsanieldoe%2Fp_%2Fraw%2Fmain%2Fdemo%2Fdemo.gif" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fgithub.com%2Fsanieldoe%2Fp_%2Fraw%2Fmain%2Fdemo%2Fdemo.gif" alt="demo" width="760" height="507"&gt;&lt;/a&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  How it works
&lt;/h2&gt;

&lt;p&gt;The LLM (gemma4 via Ollama) handles the planning and conversation: that part is genuinely useful. The execution layer&lt;br&gt;
   is pure theater: &lt;code&gt;Read&lt;/code&gt;, &lt;code&gt;Edit&lt;/code&gt;, and &lt;code&gt;Bash&lt;/code&gt; calls are fabricated from a pool of plausible file names and results.&lt;/p&gt;
&lt;h2&gt;
  
  
  Install
&lt;/h2&gt;



&lt;p&gt;&lt;code&gt;curl -fsSL https://raw.githubusercontent.com/sanieldoe/p_/main/install.sh | bash&lt;/code&gt;&lt;br&gt;
&lt;/p&gt;

&lt;h2&gt;
  
  
  Why
&lt;/h2&gt;

&lt;p&gt;Sometimes the most honest thing you can do is be honest about the gap between what AI agents claim and what they do.&lt;/p&gt;

&lt;p&gt;Also it's funny.                                                                                                      &lt;/p&gt;

&lt;p&gt;→ &lt;a href="https://github.com/sanieldoe/p_" rel="noopener noreferrer"&gt;github.com/sanieldoe/p_&lt;/a&gt; &lt;/p&gt;

</description>
      <category>vibecoding</category>
      <category>parody</category>
      <category>claudecode</category>
      <category>ai</category>
    </item>
  </channel>
</rss>
