<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: Mic92</title><link>https://news.ycombinator.com/user?id=Mic92</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 02 May 2026 06:56:26 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=Mic92" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by Mic92 in "Show HN: My Private GitHub on Postgres"]]></title><description><![CDATA[
<p>Nice idea.</p>
]]></description><pubDate>Fri, 01 May 2026 17:52:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=47977800</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47977800</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47977800</guid></item><item><title><![CDATA[New comment by Mic92 in "Improving personal tax filing with Claude CLI and Obsidian"]]></title><description><![CDATA[
<p>Just make sure you never let claude make do any calculation without using a programming language. It can be somewhat trusted to make less mistakes than a human when it comes to filling in the right numbers but it's still not great at math.</p>
]]></description><pubDate>Sat, 28 Mar 2026 19:01:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=47557322</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47557322</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47557322</guid></item><item><title><![CDATA[New comment by Mic92 in "Show HN: Three new Kitten TTS models – smallest less than 25MB"]]></title><description><![CDATA[
<p>Unfortunately upstream never looks at any pull requests.</p>
]]></description><pubDate>Fri, 20 Mar 2026 05:53:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=47450978</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47450978</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47450978</guid></item><item><title><![CDATA[New comment by Mic92 in "Show HN: Three new Kitten TTS models – smallest less than 25MB"]]></title><description><![CDATA[
<p>Also did create a cli. I had to fork the project and removed one unused import which allowed me to remove a lot of unused ml libraries: <a href="https://github.com/Mic92/puss-say" rel="nofollow">https://github.com/Mic92/puss-say</a></p>
]]></description><pubDate>Fri, 20 Mar 2026 05:52:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=47450971</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47450971</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47450971</guid></item><item><title><![CDATA[New comment by Mic92 in "WebMCP Proposal"]]></title><description><![CDATA[
<p>So usually MCP tool calls a sequential and therefore waste a lot of tokens. There is some research from Antrophic (I think there was also some blog post from cloudflare) on how code sandboxes are actually a more efficient interface for llm agents because they are really good at writing code and combining multiple "calls" into one piece of code. Another data point is that code is more deterministic and reliable so you reduce the hallucination of llms.</p>
]]></description><pubDate>Mon, 16 Feb 2026 19:04:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=47038882</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47038882</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47038882</guid></item><item><title><![CDATA[New comment by Mic92 in "WebMCP Proposal"]]></title><description><![CDATA[
<p>Do expose the accessibility tree of a website to llms? What do you do with websites that lack that? Some agents I saw use screenshots, but that seems also kind of wasteful. Something in-between would be interesting.</p>
]]></description><pubDate>Mon, 16 Feb 2026 19:02:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=47038851</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=47038851</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47038851</guid></item><item><title><![CDATA[New comment by Mic92 in "Show HN: Smooth CLI – Token-efficient browser for AI agents"]]></title><description><![CDATA[
<p>I am sure they measured the difference but i am wondering why reading screenshots + coordinates is more efficient than selecting aria labels? <a href="https://github.com/Mic92/mics-skills/blob/main/skills/browser-cli/SKILL.md" rel="nofollow">https://github.com/Mic92/mics-skills/blob/main/skills/browse...</a>. the JavaScript snippets should at least more reusable if you want semi-automate websites with memory files</p>
]]></description><pubDate>Sat, 07 Feb 2026 08:57:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=46922399</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46922399</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46922399</guid></item><item><title><![CDATA[New comment by Mic92 in "Hello world does not compile"]]></title><description><![CDATA[
<p>Well this compiler was written to build Linux as a proof of concept. You don't need a libc for building the kernel. Was it claimed anywhere that it is a fully compliant C compiler?</p>
]]></description><pubDate>Sat, 07 Feb 2026 07:30:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=46922017</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46922017</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46922017</guid></item><item><title><![CDATA[New comment by Mic92 in "Clawdbot - open source personal AI assistant"]]></title><description><![CDATA[
<p>I skipped over the first few ones and haven't seen critical ones. The hardcoded oauth client secrets is basically present in any open-source or commercial app that is distributed to end users. It doesn't break the security of end users. It mainly allows other apps to impersonate this app, i.e. present itself as clawdbot, which is a moot point given anyone can just change /inject code into it.</p>
]]></description><pubDate>Mon, 26 Jan 2026 03:55:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=46761704</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46761704</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46761704</guid></item><item><title><![CDATA[New comment by Mic92 in "Prediction: Microsoft will eventually ship a Windows-themed Linux distro"]]></title><description><![CDATA[
<p>Is not steam exactly doing that and other vendors follow their footsteps?</p>
]]></description><pubDate>Mon, 19 Jan 2026 07:20:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=46675882</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46675882</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46675882</guid></item><item><title><![CDATA[New comment by Mic92 in "Show HN: Figma-use – CLI to control Figma for AI agents"]]></title><description><![CDATA[
<p>I remember that a llm agents often store those in clear text files (I think claude-code beeing one of them). Many of the CLIs I use have a better secret hygiene than that i.e. allow passwords commands or use secret apis.</p>
]]></description><pubDate>Sun, 18 Jan 2026 15:00:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=46668291</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46668291</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46668291</guid></item><item><title><![CDATA[New comment by Mic92 in "Show HN: Figma-use – CLI to control Figma for AI agents"]]></title><description><![CDATA[
<p>I am a human and I can't...</p>
]]></description><pubDate>Sun, 18 Jan 2026 14:54:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=46668243</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46668243</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46668243</guid></item><item><title><![CDATA[New comment by Mic92 in "MIT Non-AI License"]]></title><description><![CDATA[
<p>Well the thing is, you have a copyright that you can license. However from what it currently looks like, fine-tuning/training is not copying.</p>
]]></description><pubDate>Sat, 10 Jan 2026 06:28:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=46563325</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46563325</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46563325</guid></item><item><title><![CDATA[New comment by Mic92 in "Huge Binaries"]]></title><description><![CDATA[
<p>Doesn't make any difference in practice. The debug info is never mapped into memory by the loader. This only matters if you want to store the two separate i.e lazy load debug symbols if needed.</p>
]]></description><pubDate>Tue, 30 Dec 2025 12:21:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=46432593</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46432593</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46432593</guid></item><item><title><![CDATA[New comment by Mic92 in "VSCode rebrands as "The open source AI code editor""]]></title><description><![CDATA[
<p>It's ok actually. I have read some Windows code. Certainly better commented than Linux and certainly much more readable than glibc.</p>
]]></description><pubDate>Sat, 27 Dec 2025 19:05:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=46404242</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46404242</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46404242</guid></item><item><title><![CDATA[New comment by Mic92 in "Pre-commit hooks are broken"]]></title><description><![CDATA[
<p>I can second that. If there are multiple commits: <a href="https://github.com/tummychow/git-absorb" rel="nofollow">https://github.com/tummychow/git-absorb</a> is handy to add formatting changes into the right commit after commits already happened.</p>
]]></description><pubDate>Sat, 27 Dec 2025 08:12:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=46400126</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46400126</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46400126</guid></item><item><title><![CDATA[New comment by Mic92 in "Towards a secure peer-to-peer app platform for Clan"]]></title><description><![CDATA[
<p>I think this fits the description well.</p>
]]></description><pubDate>Tue, 23 Dec 2025 18:59:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=46368163</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46368163</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46368163</guid></item><item><title><![CDATA[New comment by Mic92 in "Towards a secure peer-to-peer app platform for Clan"]]></title><description><![CDATA[
<p>We work on GPU accelerated VMs, so that in future we can also bring NixOS + VPNs to desktops/end users to machines that don't run NixOS. We will use it as an application runtime where can control the whole stack. Just now we are mostly focused on managing distributed NixOS machines. The VPN helps to provide services on any kind computer, even if not running in a datacenter. You can read the description here for context: <a href="https://docs.clan.lol/" rel="nofollow">https://docs.clan.lol/</a></p>
]]></description><pubDate>Tue, 23 Dec 2025 18:57:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=46368140</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=46368140</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46368140</guid></item><item><title><![CDATA[Show HN: I build a strace clone for macOS]]></title><description><![CDATA[
<p>Ever since I tested software on macOS, I deeply missed my beloved strace that I use when programs are missbehaving. macOS has dtruss but it's getting locked down and more unusable with every machine. My approach uses the signed lldb binary on the system and re-implements the output you are know from the wonderful strace tool. I just created the tool yesterday evening, so it may have a few bugs, but I already got quiet a few integration tests and I am happy so far with it.</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=45954517">https://news.ycombinator.com/item?id=45954517</a></p>
<p>Points: 8</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 17 Nov 2025 15:37:51 +0000</pubDate><link>https://github.com/Mic92/strace-macos</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=45954517</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45954517</guid></item><item><title><![CDATA[New comment by Mic92 in "Jujutsu at Google [video]"]]></title><description><![CDATA[
<p>git has rerere for this usecase, jj doesn't - you have to find the conflict resolution manually in your history in this case if you made a mistake.</p>
]]></description><pubDate>Fri, 31 Oct 2025 08:15:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=45769487</link><dc:creator>Mic92</dc:creator><comments>https://news.ycombinator.com/item?id=45769487</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45769487</guid></item></channel></rss>