<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Abhishek Agarwal on Medium]]></title>
        <description><![CDATA[Stories by Abhishek Agarwal on Medium]]></description>
        <link>https://medium.com/@noob-programmer?source=rss-40cfe0a41d51------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*BOy8V9VoIyhsclQfNm1nDA.gif</url>
            <title>Stories by Abhishek Agarwal on Medium</title>
            <link>https://medium.com/@noob-programmer?source=rss-40cfe0a41d51------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:03:22 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@noob-programmer/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Anthropic Just Gave Claude Code Sessions a Group Chat. The 4 Rules Behind It Are the Real Story.]]></title>
            <link>https://levelup.gitconnected.com/claude-code-cross-session-messaging-4-security-rules-16920afebccb?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/16920afebccb</guid>
            <category><![CDATA[multi-agent-systems]]></category>
            <category><![CDATA[anthropic-claude]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[claude-code]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 14:58:51 GMT</pubDate>
            <atom:updated>2026-11-06T15:14:52.079Z</atom:updated>
            <content:encoded><![CDATA[<h4>Claude Code’s new cross-session messaging: ListAgents, SendMessage, a private Unix socket, and one design rule that shows where multi-agent coding is heading.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*LXwObyowWNzhq4MM" /><figcaption>Photo by <a href="https://unsplash.com/@coopery?utm_source=medium&amp;utm_medium=referral">Mohamed Nohassi</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Right now, as I write this, I have <strong>two Claude Code sessions open on my Mac</strong>. One of them has been grinding on my app for four days straight. The other one is helping me write this post.</p><p>Until recently, those two sessions had no idea the other existed. If the app session discovered something the writing session needed, I was the bridge. I copied, I pasted, I re-explained. <strong>I was a human clipboard with a coffee habit.</strong></p><p>So I asked the writing session a simple question: who else is out there? It answered in under a second:</p><pre>This session is documents-5a (the name other sessions use to message it).<br><br>Peer sessions (1):<br>  dew-8b  ·  interactive  ·  busy  ·  started 4d ago</pre><p>That is a Claude Code session finding another Claude Code session on its own. No MCP server, no tmux hack, no shared markdown inbox. <strong>It is built in, it is on by default, and most developers I know still have not noticed it.</strong></p><h3>What Anthropic actually shipped</h3><p>Anthropic calls it <strong>cross-session messaging</strong>. It landed in Claude Code <strong>v2.1.224</strong> during the week of August 3 to 7, 2026, and has been refined in almost every release since. The whole feature is two tools:</p><p><strong>ListAgents</strong> discovers every session Claude can reach: subagents, agent-team teammates, your other local sessions, and (with Remote Control) your sessions on other machines and in the cloud.</p><p><strong>SendMessage</strong> delivers a plain-text message to one of them by name.</p><p>Here is the part people miss: <strong>you never call either tool yourself.</strong> You just talk. “Ask the session in my other terminal whether the migration finished.” Claude finds the target, writes the message, and sends it. It can also decide to send one on its own, for example right after it changes something another session depends on.</p><p>The upgrades since launch matter too. Version 2.1.232 added mentions, so you can type an at sign plus a session name (say, api-worker) and pick it from a typeahead, the same way you mention a subagent. Version 2.1.236 added <strong>notify_when_idle</strong>, which lets one session ask another for a single ping when it finishes, with no polling and no tokens spent while it waits. Native Windows support arrived in v2.1.234. To see who your session can reach, type <strong>/list-agents</strong> (or its alias /peers).</p><h3>How it works under the hood</h3><p>Every session with messaging enabled <strong>binds its own inbox</strong>: a Unix domain socket on macOS and Linux, a named pipe on native Windows. You can find yours in the Peer address row of /status, and Claude Code exports the path to your hooks and Bash commands as CLAUDE_CODE_MESSAGING_SOCKET.</p><p>I checked mine. The socket sits in a private directory, and both the directory and the socket file are locked to my user account, with no group or world access at all. Another user on the same machine cannot post into my sessions.</p><p>Where the message travels depends on where the other session lives. On the same machine it goes over that local socket and <strong>never touches Anthropic’s servers</strong>. To a session on another of your machines, it goes through Anthropic’s servers and arrives over that machine’s Remote Control connection. To a cloud session, it goes through Anthropic’s servers directly.</p><p>Delivery is polite. A busy session reads the message between tool calls, so <strong>a running test or build is never interrupted</strong>. An idle session simply starts a new turn with the message. On screen you see a dim one-line preview with the sender’s name and the first line of the message, like “Message from api-worker: Schema migration finished”. Ctrl+O expands it.</p><p>And the most important constraint: a message is text that one Claude writes for another. <strong>It never carries your conversation history or your files.</strong> If you want to move a whole conversation, Anthropic tells you to resume the session instead.</p><h3>The 4 rules that are the real story</h3><p>Anyone can wire two chatbots together. What makes this feature interesting is <strong>what Anthropic refused to let a message do</strong>. When session A messages session B, Claude Code tells B that the text came from another session, not from you, and four hard rules kick in:</p><p><strong>Rule 1: It cannot approve anything.</strong> A message from another session never counts as your consent, so it cannot answer a pending permission prompt on your behalf.</p><p><strong>Rule 2: It cannot change configuration.</strong> The receiving Claude is instructed never to edit permission settings, CLAUDE.md, or any other config just because another session asked.</p><p><strong>Rule 3: Commands do not run.</strong> A slash command inside a message, like /compact, arrives as plain text. Claude Code never executes it.</p><p><strong>Rule 4: Permission prompts still fire.</strong> If acting on the message needs a permission the receiving session does not have, you see the exact same prompt you would see for any other work.</p><blockquote>A message is information, never authority.</blockquote><p>That one sentence is the entire security model. And there is a clever asymmetry on top of it. Claude Code sorts sessions into two classes: ones that bypass permission prompts and ones that still ask. A normal session accepts messages freely, but holds anything sent from a bypassing session for your approval. A bypassing session holds everything, unless the sender also bypasses. In other words, <strong>a message can never borrow the authority of a more permissive session.</strong></p><p>There is also loop protection, because two agents politely thanking each other forever is a very real failure mode. The receiver rate limits repeated messages per sender, drops identical repeats that arrive in a short window, and queues at most 50 unread messages. Messages over roughly a million characters are refused before they leave. <strong>A runaway loop between two sessions stops on its own.</strong></p><p>If you want to be stricter, two settings do the work. crossSessionInbound can be accept, hold, or refuse. isolatePeerMachines forces an approval before any message leaves your machine, even in bypass mode. For a sensitive session, its settings can combine both:</p><pre>// .claude/settings.local.json for a sensitive session<br>{<br>  &quot;isolatePeerMachines&quot;: true,<br>  &quot;crossSessionInbound&quot;: &quot;hold&quot;<br>}</pre><p>Because a true value for isolatePeerMachines from any settings scope wins, <strong>a checked-in project file can tighten this rule but can never loosen it.</strong> That is the kind of detail that tells you a security team was in the room.</p><h3>What developers are actually saying</h3><p>The <a href="https://news.ycombinator.com/item?id=49222824">Hacker News thread</a> for the launch hit <strong>150 points and 67 comments</strong>, and the reaction was split right down the middle.</p><p>The skeptics went straight for security. One commenter wrote: “There was a time when you call this a Remote Code Execution vuln. It’s of course a feature here.” Another raised the sharper point that Remote Control can reach any session on your account, which in their words “totally breaks sandbox / VM isolation.”</p><p>The other half of the thread was a confession booth. Developers admitted they had already hacked this together with tmux send-keys, IRC servers, Telegram bots, Slack webhooks, and shared markdown files that agents take turns appending to. Some pointed out that OpenAI’s Codex already had cross-thread messaging. The demand was obviously there long before Anthropic shipped anything.</p><p>The most thoughtful critique I found came from a breakdown on <a href="https://www.versaedits.com/article/claude-sessions-can-text-other-heres-actually-means">VersaEdits</a>. Its point: once Claude writes the summary, <strong>you are no longer the filter for what crosses between sessions.</strong> A dropped detail can send the receiving session off on incomplete information. A finding from a cheap, fast model looks exactly as authoritative as one from a careful, expensive one. And every delivered message triggers a full turn in the receiver, which counts toward your usage like a prompt you typed.</p><h3>Where it fits in Claude Code’s agent zoo</h3><p>Claude Code now has a lot of ways to run agents in parallel, and this feature is easy to confuse with the others. Anthropic’s own <a href="https://code.claude.com/docs/en/agents">comparison</a> makes it clear:</p><p><strong>Subagents</strong> work inside one session and report back only to the conversation that spawned them.</p><p><strong>Agent view</strong> dispatches background sessions that report only to you.</p><p><strong>Agent teams</strong> have a lead, a shared task list, and structured messages. They are still experimental and off by default.</p><p><strong>Dynamic workflows</strong> run a script that fans out many subagents and cross-checks their results.</p><p>Cross-session messaging is the glue for sessions you run yourself. <strong>It is not a framework. It is a mailbox.</strong></p><h3>My take: the human clipboard job is gone, and that is both the win and the risk</h3><p>I think this is one of the most underrated releases Anthropic has shipped this year, and not because it is flashy. It is the opposite. <strong>It is plumbing.</strong> But plumbing is exactly what the next phase of AI coding needs.</p><p>The industry conversation has moved from prompt engineering to context engineering to what people now call loop engineering: designing the systems that keep agents working, instead of typing at them one prompt at a time. <strong>You cannot build loops out of agents that cannot talk to each other.</strong> Cross-session messaging is the missing wire between them.</p><p>But I would not hand it the keys blindly. Here is how I am setting it up:</p><p><strong>Name every important session with /rename.</strong> Auto-generated names are fine for two terminals and useless for six.</p><p><strong>Tell your sessions to treat incoming messages as claims to verify, not facts.</strong> The design already strips messages of authority; your prompts should strip them of blind trust too.</p><p><strong>Put sensitive sessions on hold.</strong> Anything touching production, credentials, or a bypass-mode run should require you to approve each inbound message.</p><p><strong>Turn on isolatePeerMachines if you use Remote Control.</strong> Local messages never leave your machine. Cross-machine ones go through Anthropic’s servers, and you should decide that consciously.</p><p><strong>Use notify_when_idle instead of babysitting.</strong> “Tell me when the migration session finishes” is, for me, the most useful prompt this feature unlocks.</p><p>The headline says your Claude Code sessions can talk to each other. The real story is that Anthropic decided, before anything else, what they are not allowed to say. <strong>Agents that can talk but cannot command each other</strong> is a pretty good blueprint for every multi-agent system we are about to build.</p><p><strong>Now I am curious: what is the first conversation you would let your agents have without you in the middle?</strong></p><p>Further reading: Anthropic’s <a href="https://code.claude.com/docs/en/cross-session-messaging">official cross-session messaging docs</a>, the <a href="https://code.claude.com/docs/en/whats-new/2026-w32">week 32 release notes</a>, the <a href="https://news.ycombinator.com/item?id=49222824">Hacker News launch thread</a>, and <a href="https://blakecrosley.com/blog/claude-code-cross-session-messaging">Blake Crosley’s deep dive</a> on the security model.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=16920afebccb" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/claude-code-cross-session-messaging-4-security-rules-16920afebccb">Anthropic Just Gave Claude Code Sessions a Group Chat. The 4 Rules Behind It Are the Real Story.</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Inside Scion: How Anthropic Replaced Git in 5 Months Using Claude]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/anthropic-scion-replaces-git-f42317d62889?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/1024/1*Mqrh81KFf77XQnPd8pPdkw.png" width="1024"></a></p><p class="medium-feed-snippet">Inside Scion, the git replacement Anthropic built in five months, and the one line in the post that matters more than any benchmark.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/anthropic-scion-replaces-git-f42317d62889?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/anthropic-scion-replaces-git-f42317d62889?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/f42317d62889</guid>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[anthropics]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[github]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Fri, 16 Oct 2026 06:03:33 GMT</pubDate>
            <atom:updated>2026-10-16T14:10:25.910Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Behind the ARTEMIS Hype: The Open-Source Scandal Google Tried to Bury]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/google-artemis-copied-minitap-open-source-ai-agent-71fda88a903e?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/2600/0*toNONRiC9sgOslGx" width="4032"></a></p><p class="medium-feed-snippet">ARTEMIS plugs straight into Claude Code and could change mobile QA forever. The story behind the repo is the part every developer should&#x2026;</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/google-artemis-copied-minitap-open-source-ai-agent-71fda88a903e?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/google-artemis-copied-minitap-open-source-ai-agent-71fda88a903e?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/71fda88a903e</guid>
            <category><![CDATA[android]]></category>
            <category><![CDATA[open-source]]></category>
            <category><![CDATA[claude-code]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[google]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Sun, 04 Oct 2026 12:34:25 GMT</pubDate>
            <atom:updated>2026-10-05T14:10:21.828Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[GitHub Is Quietly Building the Zero-Human Company. 175,000 Developers Already Joined. Have You?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/github-zero-human-company-orca-paperclip-ai-agents-10924ae5bcd0?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/1024/1*_lmHQKyhotbwhl7Hmbes1A.png" width="1024"></a></p><p class="medium-feed-snippet">Orca, Paperclip and Hindsight are turning developers into managers of AI agent fleets. The same month, OpenAI&#x2019;s agents showed what happens&#x2026;</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/github-zero-human-company-orca-paperclip-ai-agents-10924ae5bcd0?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/github-zero-human-company-orca-paperclip-ai-agents-10924ae5bcd0?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/10924ae5bcd0</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Tue, 29 Sep 2026 19:01:24 GMT</pubDate>
            <atom:updated>2026-09-30T14:40:42.186Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Anthropic Froze the Biggest IPO Ever Over Leaks, Mutiny, and One Bill]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/anthropic-ipo-delayed-november-2026-876517798d9f?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/1024/1*9q29NvVl95EoOeVq64MKJQ.png" width="1024"></a></p><p class="medium-feed-snippet">Leaked models, resigning scientists, and the single $1.25 billion bill that forced Anthropic to delay its $2 trillion debut.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/anthropic-ipo-delayed-november-2026-876517798d9f?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/anthropic-ipo-delayed-november-2026-876517798d9f?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/876517798d9f</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[anthropics]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[stock-market]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Wed, 23 Sep 2026 09:36:19 GMT</pubDate>
            <atom:updated>2026-09-23T14:11:40.455Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Meet Jev: The ChatGPT Co-Creator’s AI That Can’t Even Say Hello (and It’s 100x Faster)]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/jev-ai-explained-typesafe-system-one-model-b0292bab4ac3?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/1024/1*C2FeKbHmHuiIQIIhCXKMjQ.png" width="1024"></a></p><p class="medium-feed-snippet">Everything you need to know about TypeSafe&#x2019;s Jev: how it works, what it costs, what Hacker News tore apart, and where I&#x2019;d actually use it.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/jev-ai-explained-typesafe-system-one-model-b0292bab4ac3?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/jev-ai-explained-typesafe-system-one-model-b0292bab4ac3?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/b0292bab4ac3</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[llm]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Mon, 21 Sep 2026 15:24:24 GMT</pubDate>
            <atom:updated>2026-09-23T14:15:45.285Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Anthropic Wants a Selfie Before You Use Claude. Blame 24,000 Fake Accounts]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/claude-age-verification-yoti-persona-c819fed9a76f?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/1024/1*Lm4iZdI7HxXh2kHrzOIMQQ.png" width="1024"></a></p><p class="medium-feed-snippet">Yoti guesses your age from a selfie. Persona wants your passport. And 24,000 fake accounts explain why your chatbot is starting to act&#x2026;</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/claude-age-verification-yoti-persona-c819fed9a76f?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/claude-age-verification-yoti-persona-c819fed9a76f?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/c819fed9a76f</guid>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[anthropics]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 19:56:26 GMT</pubDate>
            <atom:updated>2026-09-14T14:49:51.491Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Anthropic Just Cut Claude Code’s Limits by 17% and Announced It as a 25% Increase]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/anthropic-cut-claude-code-limits-17-percent-2af51725b0a1?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/2600/0*Onm1g3xQezGtuXyh" width="4104"></a></p><p class="medium-feed-snippet">Both numbers are true. Only one of them is what happens to you on September 14.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/anthropic-cut-claude-code-limits-17-percent-2af51725b0a1?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/anthropic-cut-claude-code-limits-17-percent-2af51725b0a1?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/2af51725b0a1</guid>
            <category><![CDATA[anthropics]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Tue, 08 Sep 2026 13:06:35 GMT</pubDate>
            <atom:updated>2026-09-09T14:23:46.273Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[You’re Not Bad at Prompting. You’re Missing the 9 Things This Claude Skill Checks For.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/prompt-master-claude-skill-any-ai-264e4949d2f3?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/2600/0*btIWZigw2V4PfDav" width="6000"></a></p><p class="medium-feed-snippet">12,126 stars in five months. Here&#x2019;s the extraction pipeline &#x2014; and what it actually costs you.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/prompt-master-claude-skill-any-ai-264e4949d2f3?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/prompt-master-claude-skill-any-ai-264e4949d2f3?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/264e4949d2f3</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[claude]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Sun, 06 Sep 2026 04:59:25 GMT</pubDate>
            <atom:updated>2026-09-07T15:18:41.204Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Don’t Fall for OpenAI’s 99.9% AGI Lie. Anthropic Just Won the War.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/gpt-6-astra-vs-claude-fable-5-1-agi-era-cf2b5707de0d?source=rss-40cfe0a41d51------2"><img src="https://cdn-images-1.medium.com/max/2600/0*2uyO0BE_gv75AVkH" width="6000"></a></p><p class="medium-feed-snippet">Two days earlier, with no press release. And that 99.9% everyone quoted isn&#x2019;t the number ARC Prize actually measured.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/gpt-6-astra-vs-claude-fable-5-1-agi-era-cf2b5707de0d?source=rss-40cfe0a41d51------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/gpt-6-astra-vs-claude-fable-5-1-agi-era-cf2b5707de0d?source=rss-40cfe0a41d51------2</link>
            <guid isPermaLink="false">https://medium.com/p/cf2b5707de0d</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[anthropics]]></category>
            <dc:creator><![CDATA[Abhishek Agarwal]]></dc:creator>
            <pubDate>Fri, 04 Sep 2026 13:04:45 GMT</pubDate>
            <atom:updated>2026-09-04T14:04:33.821Z</atom:updated>
        </item>
    </channel>
</rss>