<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: heysrb</title><link>https://news.ycombinator.com/user?id=heysrb</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 19:17:23 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=heysrb" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by heysrb in "Show HN: HackerTV – HN's most-recommended Youtube videos as 24/7 TV channels"]]></title><description><![CDATA[
<p>This is cooler than Karpathy's!</p>
]]></description><pubDate>Wed, 04 Nov 2026 15:44:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=50367213</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50367213</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50367213</guid></item><item><title><![CDATA[Ask HN: What is your best hack to keep your agent's context in sync?]]></title><description><![CDATA[
<p>A lot of people manually maintain markdown files and use them to sync context. Some people even use Obsidian.<p>I was wondering what is the best hack you have discovered to keep the agents in sync? And how does it perform, does it work reliably well?<p>One of the problems I face is multiple agents writing to a common memory which can lead to unwanted memory overwrites without any trace.</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50305638">https://news.ycombinator.com/item?id=50305638</a></p>
<p>Points: 1</p>
<p># Comments: 1</p>
]]></description><pubDate>Fri, 30 Oct 2026 15:48:14 +0000</pubDate><link>https://news.ycombinator.com/item?id=50305638</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50305638</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50305638</guid></item><item><title><![CDATA[New comment by heysrb in "Show HN: Give one context across all your agents via MCP"]]></title><description><![CDATA[
<p>If you ever thought "What if I could just continue my session on Codex" after exhausting your Claude limits then Make0 AI will work out for you.<p>Make0 is common a memory store for your agents. Claude, Codex, Gemini all read and write to this common memory store and stay in sync with common context.<p>There is memory versioning so that you can see which agent did what and can easily restore. Do give it a shot!</p>
]]></description><pubDate>Mon, 26 Oct 2026 15:44:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=50245705</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50245705</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50245705</guid></item><item><title><![CDATA[Show HN: Give one context across all your agents via MCP]]></title><description><![CDATA[
<p>Article URL: <a href="https://make0.ai/">https://make0.ai/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50245585">https://news.ycombinator.com/item?id=50245585</a></p>
<p>Points: 1</p>
<p># Comments: 1</p>
]]></description><pubDate>Mon, 26 Oct 2026 15:38:33 +0000</pubDate><link>https://make0.ai/</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50245585</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50245585</guid></item><item><title><![CDATA[New comment by heysrb in "Show HN: Working on Plug and play AI memory that works across AI agents"]]></title><description><![CDATA[
<p>LoCoMo has an open world question category. Which is non factual/subjective retrieval (96 questions). Like "Based on everything you know about Maria, what birthday gift would she appreciate?"<p>These are challenging as the retrievers need understanding not just keyword matching or vector similarity.<p>Although subset of questions is less but a lot of systems fail at this.</p>
]]></description><pubDate>Mon, 26 Oct 2026 15:34:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=50245510</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50245510</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50245510</guid></item><item><title><![CDATA[Show HN: Working on Plug and play AI memory that works across AI agents]]></title><description><![CDATA[
<p>So I've been working on a project (Make0 AI) for a long time where I wanted to give common context to both Codex and Claude Code. The problem is simple - while working on Claude and after exhausting the daily/weekly limits I've start again from scratch in Codex. Of course I can prompt Claude to summarise a handoff that I can give to Codex but that is inefficient.
So I thought of a creating a common context for both of them locally using a simple list of instructions like a shared md file. But the problem was the file kept on growing and after some point the common context went beyond 20k tokens and was just inflating the token budget unnecessarily.<p>Then I implemented a RAG based retrieval approach on the list to only load what is required for that prompt and voila, after a few more iterations it resulted into a graph memory - Brain.<p>So this product Make0 AI gives you the ability to create Brains that you can connect to any Ai agent via MCP and then they immediately know your likes, dislikes, preferences etc. AI agents can read, write to this memory.<p>This can be your long term memory and the agents just become the token providers and task executors. You have full control over your data/memories from a single place. Think of this as a "USB thumb drive brain" that you plug to any agent.<p>I know I may not be the first one to implement this but I tried to make the UX simple and is not at all developer focused. You can just drag and drop your memories/data to the brain and even play around with a visual brain.<p>One more thing, i've done benchmarks on the memory. LoCoMo scores 83.3% overall, with Single hop 92.5% in standard mode (available now) and in another configuration (experimental max mode) it scores 92% overall and 94.4% in single hop. Available in the benchmark page. It uses hybrid search + graph memory and has built in conflict resolution for memories.<p>Do let me know if you folks wanna try it!</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50100700">https://news.ycombinator.com/item?id=50100700</a></p>
<p>Points: 1</p>
<p># Comments: 2</p>
]]></description><pubDate>Thu, 15 Oct 2026 08:02:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=50100700</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=50100700</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50100700</guid></item><item><title><![CDATA[Working on Plug and Play personal AI Memory that works across AI agents]]></title><description><![CDATA[
<p>So I've been working on a project (Make0 AI) for a long time where I wanted to give common context to both Codex and Claude Code. The problem is simple - while working on Claude and after exhausting the daily/weekly limits I've start again from scratch in Codex. Of course I can prompt Claude to summarise a handoff that I can give to Codex but that is inefficient.<p>So I thought of a creating a common context for both of them locally using a simple list of instructions like a shared md file. But the problem was the file kept on growing and after some point the common context went beyond 20k tokens and was just inflating the token budget unnecessarily.<p>Then I implemented a RAG based retrieval approach on the list to only load what is required for that prompt and voila, after a few more iterations it resulted into a graph memory - Brain.<p>So this product Make0 AI gives you the ability to create Brains that you can connect to any Ai agent via MCP and then they immediately know your likes, dislikes, preferences etc. AI agents can read, write to this memory.<p>This can be your long term memory and the agents just become the token providers and task executors. You have full control over your data/memories from a single place. Think of this as a "USB thumb drive brain" that you plug to any agent.<p>I know I may not be the first one to implement this but I tried to make the UX simple and is not at all developer focused. You can just drag and drop your memories/data to the brain and even play around with a visual brain.<p>One more thing, i've done benchmarks on the memory. LoCoMo scores 83.3% overall, with Single hop 92.5% in standard mode (available now) and in another configuration (experimental max mode) it scores 92% overall and 94.4% in single hop. Available in the benchmark page.<p>I'm currently keeping this invite only. Do let me know if you folks wanna try it!</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49713789">https://news.ycombinator.com/item?id=49713789</a></p>
<p>Points: 3</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 15 Sep 2026 15:13:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=49713789</link><dc:creator>heysrb</dc:creator><comments>https://news.ycombinator.com/item?id=49713789</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49713789</guid></item></channel></rss>