<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: tipoffdosage904</title><link>https://news.ycombinator.com/user?id=tipoffdosage904</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Thu, 16 Apr 2026 01:07:02 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=tipoffdosage904" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by tipoffdosage904 in "I de-vibed a vibe-coded NLP app"]]></title><description><![CDATA[
<p>Earlier this year I built a local app called How I Prompt that analyzes your AI coding conversations and gives you a prompt/persona breakdown.<p>The first version was very much vibe-coded. It looked good, but once I audited it properly I found several issues:<p>- the behavioral axes used to generate prompting personas were heavily correlated and mostly measuring prompt length<p>- one persona was mathematically unreachable<p>- the pipeline was counting logs / tool noise / machine-generated text as human prompts (this was the highest LOE for me to clean the data myself)<p>- the whole thing had a stronger appearance of rigor than actual rigor<p>I wrote up the rebuild here: <a href="https://theasymptotic.substack.com/p/how-i-de-vibed-a-vibe-coded-nlp-app" rel="nofollow">https://theasymptotic.substack.com/p/how-i-de-vibed-a-vibe-c...</a><p>Repo: <a href="https://github.com/eeshansrivastava89/howiprompt" rel="nofollow">https://github.com/eeshansrivastava89/howiprompt</a><p>What I changed in v2:<p>- much more aggressive data cleaning<p>- simpler feature-based scoring using logistic regression instead of embeddings (something I understand better)<p>- external prompt datasets for broader validation<p>- a more transparent 2-axis system that seems to behave much better than the original<p>It runs locally and doesn't upload your prompt data anywhere. Point your agent at the repo to validate yourself.<p>Would especially love feedback from people who have worked on behavioral measurement, NLP evaluation, or human/AI interaction. I'm definitely not a domain expert. One of the main things I wanted to document here was the difference between "AI helped me ship a prototype fast" and "this is actually a sound measurement system."</p>
]]></description><pubDate>Mon, 06 Apr 2026 15:07:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=47661907</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47661907</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47661907</guid></item><item><title><![CDATA[I de-vibed a vibe-coded NLP app]]></title><description><![CDATA[
<p>Article URL: <a href="https://theasymptotic.substack.com/p/how-i-de-vibed-a-vibe-coded-nlp-app">https://theasymptotic.substack.com/p/how-i-de-vibed-a-vibe-coded-nlp-app</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47661906">https://news.ycombinator.com/item?id=47661906</a></p>
<p>Points: 2</p>
<p># Comments: 2</p>
]]></description><pubDate>Mon, 06 Apr 2026 15:07:00 +0000</pubDate><link>https://theasymptotic.substack.com/p/how-i-de-vibed-a-vibe-coded-nlp-app</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47661906</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47661906</guid></item><item><title><![CDATA[New comment by tipoffdosage904 in "Claude Code Unpacked : A visual guide"]]></title><description><![CDATA[
<p>This is actually a GREAT example of vibe-coding + skill.</p>
]]></description><pubDate>Thu, 02 Apr 2026 01:23:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=47608909</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47608909</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47608909</guid></item><item><title><![CDATA[New comment by tipoffdosage904 in "Agentic Coding for Non-Vibe Coders"]]></title><description><![CDATA[
<p>This is fantastic! Thanks for sharing. Great step forward towards a pragmatic approach to truly leveraging agentic coding to increase productivity and not slop. Love this.</p>
]]></description><pubDate>Thu, 12 Mar 2026 04:45:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=47346560</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47346560</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47346560</guid></item><item><title><![CDATA[New comment by tipoffdosage904 in "Agentic Coding for Non-Vibe Coders"]]></title><description><![CDATA[
<p>My workflow for building side projects and work tools with AI coding agents that actually survive past the first month. Covers model choices (Claude Opus, Codex, Qwen), a docs-first approach (PITCH/ARCHITECTURE/IMPLEMENTATION), guardrails, context management via slash commands, and what I stopped using (MCP servers, multi-agent teams, instruction files). Includes dogfooding results — what shipped and what broke.</p>
]]></description><pubDate>Sun, 08 Mar 2026 06:06:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=47294957</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47294957</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47294957</guid></item><item><title><![CDATA[Agentic Coding for Non-Vibe Coders]]></title><description><![CDATA[
<p>Article URL: <a href="https://theasymptotic.substack.com/p/agentic-coding-for-non-vibe-coders">https://theasymptotic.substack.com/p/agentic-coding-for-non-vibe-coders</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47294956">https://news.ycombinator.com/item?id=47294956</a></p>
<p>Points: 3</p>
<p># Comments: 4</p>
]]></description><pubDate>Sun, 08 Mar 2026 06:06:18 +0000</pubDate><link>https://theasymptotic.substack.com/p/agentic-coding-for-non-vibe-coders</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47294956</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47294956</guid></item><item><title><![CDATA[The Dopamine Trap of AI Coding]]></title><description><![CDATA[
<p>Article URL: <a href="https://theasymptotic.substack.com/p/the-dopamine-trap-of-ai-coding">https://theasymptotic.substack.com/p/the-dopamine-trap-of-ai-coding</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47124470">https://news.ycombinator.com/item?id=47124470</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 23 Feb 2026 16:24:26 +0000</pubDate><link>https://theasymptotic.substack.com/p/the-dopamine-trap-of-ai-coding</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=47124470</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47124470</guid></item><item><title><![CDATA[New comment by tipoffdosage904 in "The FTC wants to ban tough-to-cancel subscriptions"]]></title><description><![CDATA[
<p>I hope this includes gym memberships!</p>
]]></description><pubDate>Fri, 24 Mar 2023 05:51:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=35285514</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=35285514</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=35285514</guid></item><item><title><![CDATA[New comment by tipoffdosage904 in "The FTC wants to ban tough-to-cancel subscriptions"]]></title><description><![CDATA[
<p>I hope this included gym memberships!</p>
]]></description><pubDate>Fri, 24 Mar 2023 05:51:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=35285511</link><dc:creator>tipoffdosage904</dc:creator><comments>https://news.ycombinator.com/item?id=35285511</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=35285511</guid></item></channel></rss>