<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Sakshyam Pokhrel — AI Product Builder &amp; Founder</title>
    <link>https://pokhrel.tech/</link>
    <description>Sakshyam Pokhrel (@pokhrelboss) is an AI product builder, developer, and founder of CraftResearch Tech—building CraftCode and ambitious software systems.</description>
    <language>en-US</language>
    <lastBuildDate>Sat, 15 Aug 2026 18:16:19 GMT</lastBuildDate>
    <atom:link href="https://pokhrel.tech/rss.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>CraftCode Architecture: Building a Deterministic AI-Native Development Harness</title>
      <link>https://pokhrel.tech/blog/craftcode-architecture-lessons/</link>
      <guid isPermaLink="true">https://pokhrel.tech/blog/craftcode-architecture-lessons/</guid>
      <description>Lessons from building CraftCode's multi-agent execution harness: deterministic tool orchestration, sandboxed runtime isolation, state recovery, and avoiding brittle autonomous loops.</description>
      <pubDate>Sat, 01 Aug 2026 10:00:00 GMT</pubDate>
      <author>sakshyam@pokhrel.tech (Sakshyam Pokhrel)</author>
      <category>Agent Systems</category>
    </item>
    <item>
      <title>Hardening Execution Sandboxes for Autonomous Coding Agents</title>
      <link>https://pokhrel.tech/blog/autonomous-coding-agents-sandbox-design/</link>
      <guid isPermaLink="true">https://pokhrel.tech/blog/autonomous-coding-agents-sandbox-design/</guid>
      <description>How to safely let LLMs execute bash, edit files, and launch dev servers without risking host compromise, infinite network loops, or accidental data loss.</description>
      <pubDate>Wed, 15 Jul 2026 09:00:00 GMT</pubDate>
      <author>sakshyam@pokhrel.tech (Sakshyam Pokhrel)</author>
      <category>Security & Infrastructure</category>
    </item>
    <item>
      <title>Benchmarking Sub-30B Coding Models on Real-World Refactoring Tasks</title>
      <link>https://pokhrel.tech/blog/local-model-benchmarking-hermes-codex/</link>
      <guid isPermaLink="true">https://pokhrel.tech/blog/local-model-benchmarking-hermes-codex/</guid>
      <description>A rigorous evaluation of small open-weight coding models on multi-file code navigation, AST manipulation, and edge-case test generation.</description>
      <pubDate>Sat, 20 Jun 2026 12:00:00 GMT</pubDate>
      <author>sakshyam@pokhrel.tech (Sakshyam Pokhrel)</author>
      <category>Benchmarks</category>
    </item>
  </channel>
</rss>