<?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: PlainSkillsZero0</title>
    <description>The latest articles on DEV Community by PlainSkillsZero0 (@plainskillszero0).</description>
    <link>https://dev.to/plainskillszero0</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%2F4215850%2F31d36739-b1d3-4860-ba38-70c9c8fbdca4.png</url>
      <title>DEV Community: PlainSkillsZero0</title>
      <link>https://dev.to/plainskillszero0</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/plainskillszero0"/>
    <language>en</language>
    <item>
      <title>Claude Code keeps forgetting? 5 fixes you can apply today</title>
      <dc:creator>PlainSkillsZero0</dc:creator>
      <pubDate>Sat, 14 Nov 2026 02:04:55 +0000</pubDate>
      <link>https://dev.to/plainskillszero0/claude-code-keeps-forgetting-5-fixes-you-can-apply-today-47hi</link>
      <guid>https://dev.to/plainskillszero0/claude-code-keeps-forgetting-5-fixes-you-can-apply-today-47hi</guid>
      <description>&lt;p&gt;If Claude Code loses the thread between sessions, you do not need a new tool. Five habits fix most of it. Everything below is generic; nothing to buy to use it.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Put what must survive in CLAUDE.md, and keep it short
&lt;/h2&gt;

&lt;p&gt;Write facts the next session cannot guess (commands, folders, "never do X"). One line each, checkable. Delete lines that stopped being true. Check the current format in the official docs: &lt;a href="https://code.claude.com/docs" rel="noopener noreferrer"&gt;https://code.claude.com/docs&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Before a long job, estimate the cost
&lt;/h2&gt;

&lt;p&gt;Write down how many steps, how many sub-agents, and what it would cost if it runs 3x longer than planned. If you cannot say, shrink the job. Use your plan's usage view if it has one.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Never accept "done" without proof
&lt;/h2&gt;

&lt;p&gt;Ask: "Check it a different way than you did it (a second command, a test, a diff) and show me the output."&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Stop loops before they burn your plan
&lt;/h2&gt;

&lt;p&gt;Every sub-task gets a one-line goal, a limit (steps or minutes) and a stop rule ("stop if two tries give the same error"). There is a small free MIT-licensed skill for this: &lt;a href="https://github.com/happy-program-zero/loop-stop" rel="noopener noreferrer"&gt;https://github.com/happy-program-zero/loop-stop&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Hand over properly at the end of a session
&lt;/h2&gt;

&lt;p&gt;Ask for 5 lines: what is done (with evidence), what is not, what is next, what is risky, which files changed. Save them as HANDOFF.md and start the next session by reading it.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;This article was drafted with AI assistance.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Independent, not affiliated with Anthropic. "Claude" and "Claude Code" are trademarks of Anthropic.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>claude</category>
      <category>productivity</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>A 15-line skill that stops Claude Code agents from burning your quota in loops</title>
      <dc:creator>PlainSkillsZero0</dc:creator>
      <pubDate>Thu, 29 Oct 2026 18:01:11 +0000</pubDate>
      <link>https://dev.to/plainskillszero0/a-15-line-skill-that-stops-claude-code-agents-from-burning-your-quota-in-loops-2ofo</link>
      <guid>https://dev.to/plainskillszero0/a-15-line-skill-that-stops-claude-code-agents-from-burning-your-quota-in-loops-2ofo</guid>
      <description>&lt;p&gt;If you let a Claude Code agent run unattended, you have probably seen it: the same failing command, retried five times, each retry eating tokens. Or a "sweep" that keeps re-checking a file that has not changed. The expensive part is not the work, it is the work that goes nowhere.&lt;/p&gt;

&lt;p&gt;We run several Claude Code sessions every day on one project, and the cheapest fix we found was not a bigger limit. It was two habits, written down once as a skill:&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Brief the agent like a colleague, not a command
&lt;/h2&gt;

&lt;p&gt;Before launching a subagent or a repeating task, write four things:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Goal and why.&lt;/strong&gt; The agent works better when it knows the purpose, not only the task.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;What it may touch.&lt;/strong&gt; Files or folders, nothing else.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A budget.&lt;/strong&gt; Turns, tokens or minutes.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The stops&lt;/strong&gt; (below).&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  2. Decide in advance when to stop
&lt;/h2&gt;

&lt;p&gt;Stop and rethink, rather than raising the limit or retrying the same way, if:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Two checks in a row show no measurable progress (same state, same file, same figure).&lt;/li&gt;
&lt;li&gt;The same error appears twice. Change the cause (a tool in a loop, an API limit) before retrying.&lt;/li&gt;
&lt;li&gt;The budget of the brief is exceeded.&lt;/li&gt;
&lt;li&gt;A git conflict or a locked file blocks progress. Do not force it.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;And when the result comes back: at most three follow-up rounds, each with a reason. On the third, close it as &lt;strong&gt;applied&lt;/strong&gt; or &lt;strong&gt;discarded&lt;/strong&gt;. Nothing stays in limbo.&lt;/p&gt;

&lt;h2&gt;
  
  
  The skill
&lt;/h2&gt;

&lt;p&gt;That is the whole thing. It fits in a plain-text &lt;code&gt;SKILL.md&lt;/code&gt; you can read in thirty seconds and drop in your skills folder:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight markdown"&gt;&lt;code&gt;&lt;span class="nn"&gt;---&lt;/span&gt;
&lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;loop-stop&lt;/span&gt;
&lt;span class="na"&gt;description&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&lt;/span&gt;&lt;span class="s"&gt;Brief&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;a&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;long-running&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;agent&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;or&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;loop&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;properly&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;and&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;know&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;when&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;to&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;stop&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;it.&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;Use&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;when&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;launching&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;or&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;watching&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;a&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;subagent,&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;a&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;sweep&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;or&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;any&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;repeating&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;task."&lt;/span&gt;
&lt;span class="nn"&gt;---&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The full file, with a short "why it exists" note, is free to use and copy for your own work: &lt;a href="https://github.com/happy-program-zero/loop-stop" rel="noopener noreferrer"&gt;https://github.com/happy-program-zero/loop-stop&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Honest notes
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;This comes from our own experience on one project. We have not measured savings across other projects, so we make no promise about results.&lt;/li&gt;
&lt;li&gt;It is plain text. Read it before you use it; there is nothing hidden in it.&lt;/li&gt;
&lt;li&gt;It is not affiliated with Anthropic. "Claude" and "Claude Code" are trademarks of Anthropic.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you want the rest of what we use (cross-checking a "done" by another route, a quota budget before long jobs, a daily hand-off report, file-based memory between sessions), there is a paid pack of seven skills: &lt;a href="https://plainskillszero0.gumroad.com/l/multi-ai-pack-complete" rel="noopener noreferrer"&gt;https://plainskillszero0.gumroad.com/l/multi-ai-pack-complete&lt;/a&gt;. It is optional; the free one stands on its own.&lt;/p&gt;

&lt;p&gt;What stop rules do you use for your own agents? I would like to hear which ones we are missing.&lt;/p&gt;

</description>
      <category>claudecode</category>
      <category>ai</category>
      <category>productivity</category>
      <category>devtools</category>
    </item>
  </channel>
</rss>
