<?xml version="1.0" encoding="UTF-8"?>
  <rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
    <channel>
      <title>Rajdeep Mondal - Claude Code, Under The Hood</title>
      <link>https://rajdeepmondal.com/series/claude-code-under-the-hood</link>
      <description>Claude Code, Under The Hood: every entry in the series.</description>
      <language>en-us</language>
      <managingEditor>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</managingEditor>
      <webMaster>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</webMaster>
      <lastBuildDate>Sun, 24 May 2026 00:00:00 GMT</lastBuildDate>
      <atom:link href="https://rajdeepmondal.com/series/claude-code-under-the-hood/feed.xml" rel="self" type="application/rss+xml"/>
      
  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-shape-coding-agents-converge-on</guid>
    <title>The Shape Coding Agents Converge On</title>
    <link>https://rajdeepmondal.com/writing/the-shape-coding-agents-converge-on</link>
    <description>Every serious coding agent is converging on one shape. The same hard limits force the same parts, whatever brand sits on the front. Build the ring around the model, because the ring is where the value stays.</description>
    
    <pubDate>Sun, 24 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/feature-gates-show-the-product-pressure</guid>
    <title>Feature Gates Show The Product Pressure</title>
    <link>https://rajdeepmondal.com/writing/feature-gates-show-the-product-pressure</link>
    <description>The switched-off corners of Claude Code reach for one kind of work. It outlives the prompt, runs far away, remembers, plans, and shares. Every corner builds from parts the earlier ideas already named. That reuse is the clue worth reading.</description>
    
    <pubDate>Fri, 22 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/everything-is-pluggable</guid>
    <title>Everything Is Pluggable</title>
    <link>https://rajdeepmondal.com/writing/everything-is-pluggable</link>
    <description>Six rings snap onto a core that never moves. Around two dozen named moments in a session hand control to your own code. Every piece carries a tag naming where it came from.</description>
    
    <pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-right-primitive</guid>
    <title>The Right Primitive</title>
    <link>https://rajdeepmondal.com/writing/the-right-primitive</link>
    <description>Five primitives carry every behavior you add to a coding agent: hook, command, skill, tool, subagent. They line up by weight. Ask the questions lightest first and take the first yes. Name that primitive in one sentence before you build.</description>
    
    <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/subagents-turn-the-cli-into-a-small-scheduler</guid>
    <title>Subagents Turn The CLI Into A Small Scheduler</title>
    <link>https://rajdeepmondal.com/writing/subagents-turn-the-cli-into-a-small-scheduler</link>
    <description>A subagent is a scoped runtime. It carries its own tools, its own permission, its own transcript, and often its own workspace. It can run in the background while you work. The scope you set before dispatch decides whether it helps.</description>
    
    <pubDate>Sat, 16 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/done-means-verified</guid>
    <title>Done Means Verified</title>
    <link>https://rajdeepmondal.com/writing/done-means-verified</link>
    <description>&quot;Done&quot; is the most expensive word an agent can say without proof. The words sound the same whether the change works or fails. Treat &quot;done&quot; as a claim until a check runs the part that changed.</description>
    
    <pubDate>Thu, 14 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-agent-is-built-to-fail-well</guid>
    <title>The Agent Is Built To Fail Well</title>
    <link>https://rajdeepmondal.com/writing/the-agent-is-built-to-fail-well</link>
    <description>Every break in a coding agent gets a named repair. A blip gets a retry, a long answer gets more room, an overflowing prompt gets trimmed. When every repair fails, the turn stops and says why. Trust lives in that clean stop.</description>
    
    <pubDate>Tue, 12 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/staying-in-control</guid>
    <title>Staying In Control</title>
    <link>https://rajdeepmondal.com/writing/staying-in-control</link>
    <description>Autonomy is worth handing over only when you can take it back. Every other control rests on one brake the agent cannot refuse. Your input lands at the next safe point between steps, so a redirect never breaks the run.</description>
    
    <pubDate>Sun, 10 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/reading-an-unfamiliar-codebase</guid>
    <title>Reading A Codebase It Has Never Seen</title>
    <link>https://rajdeepmondal.com/writing/reading-an-unfamiliar-codebase</link>
    <description>The first job in an unfamiliar codebase is to stop being lost, cheaply. Read the names, then search, then open a file. Each step costs more than the one before it, so judge an agent by how few files it opens.</description>
    
    <pubDate>Fri, 08 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/thinking-is-cheaper-than-doing</guid>
    <title>Thinking Is Cheaper Than Doing</title>
    <link>https://rajdeepmondal.com/writing/thinking-is-cheaper-than-doing</link>
    <description>Some of what an agent does is reversible and costs nothing to redo. The rest sets, and a mistake there costs a cleanup. Spend everything you can in the cheap half. Review the plan before the agent crosses.</description>
    
    <pubDate>Wed, 06 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/memory-after-the-laptop-closes</guid>
    <title>Memory After The Laptop Closes</title>
    <link>https://rajdeepmondal.com/writing/memory-after-the-laptop-closes</link>
    <description>Memory is the part I got wrong first. I reached for a bigger window. The fix was a small diary, gated at every page. Judge a memory system by what it refuses to write down, and by how it tears out a wrong note.</description>
    
    <pubDate>Mon, 04 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/why-it-feels-fast</guid>
    <title>Why It Feels Fast</title>
    <link>https://rajdeepmondal.com/writing/why-it-feels-fast</link>
    <description>A slow agent can still feel fast. You build that gap out of three moves. Stream each word as it lands, run the independent chores together. Start the next tool before the message ends. Latency is a product decision.</description>
    
    <pubDate>Sat, 02 May 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-prompt-is-an-economy</guid>
    <title>The Prompt Is An Economy</title>
    <link>https://rajdeepmondal.com/writing/the-prompt-is-an-economy</link>
    <description>The model reuses the prompt from the front and stops at the first change. Part order sets the bill and the speed. Put the steady parts first and the new message last, and most of a turn comes back free.</description>
    
    <pubDate>Thu, 30 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/context-is-managed-not-infinite</guid>
    <title>Context Is Managed, Not Infinite</title>
    <link>https://rajdeepmondal.com/writing/context-is-managed-not-infinite</link>
    <description>For a long time I wanted a bigger window. The wish was wrong. A window is a budget with a hard edge, and Claude Code defends it early. It drops duplicate reads, defers rare tools, and folds the rest into one note.</description>
    
    <pubDate>Tue, 28 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/permissions-are-a-runtime-not-a-popup</guid>
    <title>Permissions Are A Runtime, Not A Popup</title>
    <link>https://rajdeepmondal.com/writing/permissions-are-a-runtime-not-a-popup</link>
    <description>The runtime decides whether a tool runs, with a ladder of rules and then a race between answerers. Whatever the ladder does not settle falls to the race: you, a hook, an automatic check, or a phone. The first answer wins.</description>
    
    <pubDate>Sun, 26 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-tool-system-is-the-product</guid>
    <title>The Tool System Is The Product</title>
    <link>https://rajdeepmondal.com/writing/the-tool-system-is-the-product</link>
    <description>A tool is a contract the runtime checks before anything runs. The contract fixes the input, the safety, the rule that decides, and the shape of the result. It fixes what later calls can count on. The order of those checks is the product.</description>
    
    <pubDate>Fri, 24 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/the-system-prompt-is-the-agents-worldview</guid>
    <title>The System Prompt Is The Agent&#39;s Worldview</title>
    <link>https://rajdeepmondal.com/writing/the-system-prompt-is-the-agents-worldview</link>
    <description>An agent is a model plus a worldview you reinstall on every turn. That worldview is the real programming, and it arrives as one briefing card. A line earns its place on that card only by changing a decision the agent makes.</description>
    
    <pubDate>Wed, 22 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

  <item>
    <guid isPermaLink="true">https://rajdeepmondal.com/writing/claude-code-is-a-state-machine</guid>
    <title>Claude Code Is A State Machine</title>
    <link>https://rajdeepmondal.com/writing/claude-code-is-a-state-machine</link>
    <description>Claude Code is a machine that loops through model calls and tool results. It stops at one of a fixed list of endings. It reads what came back rather than the label on it. Hold that picture and the rest gets simpler.</description>
    
    <pubDate>Mon, 20 Apr 2026 00:00:00 GMT</pubDate>
    <author>rajdeep@rajdeepmondal.com (Rajdeep Mondal)</author>
    <category>Agents</category><category>Claude Code</category><category>AI Systems</category>
  </item>

    </channel>
  </rss>