<h2>Must know</h2>
<ul>
<li><strong>Anthropic source code leak exposes internal agent secrets</strong>: Anthropic accidentally leaked 512,000 lines of Claude Code TypeScript source via a misconfigured npm package and a Bun runtime bug. The leak, confirmed by multiple technical analyses, reveals unreleased features like ‘KAIROS’ for persistent autonomous agents and an ‘Undercover Mode’ that strips AI attribution from git commits. While Anthropic is aggressively issuing DMCA takedowns, the architectural blueprints for their tool-calling and memory systems are already being mirrored and rewritten by the community. <a href="https://app.daily.dev/feed-by-ids?id=9nHqJG8ou&amp;id=NKOtqwo2C&amp;id=Sfol5ptnP&amp;id=gwAJE9Fz9&amp;id=coOVHaDO4&amp;id=dKE1J9HVx&amp;id=eBpN3r6rp&amp;id=kXi9WLT2r&amp;id=cJqnnyxUG&amp;id=lSeOtUf5e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Axios and LiteLLM hit by targeted supply chain attacks</strong>: The popular Axios npm package was hijacked by North Korean threat actor UNC1069, injecting a remote access Trojan (RAT) into versions 1.14.1 and 0.30.4. Simultaneously, the LiteLLM project suffered a PyPI breach that affected high-profile users like Mercor, a $10B AI recruiting startup. Security researchers at Socket and Zscaler note these attacks specifically target developers’ AWS keys and GitHub tokens, exploiting the ‘vibe coding’ trend where AI-generated code pulls in unvetted third-party dependencies. <a href="https://app.daily.dev/feed-by-ids?id=1i5FA0Q5W&amp;id=YtEQA7dzF&amp;id=sjXZykGnw&amp;id=G3B5gEUdG&amp;id=mu01Exy9P&amp;id=vkoh5H7xM&amp;id=waOklObFm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google releases Gemma 4 with native multimodality</strong>: Google has launched Gemma 4, a family of open-weight models including a 26B MoE and a 31B dense variant, now available under Apache 2.0. These models feature native support for vision and audio with a 256K context window, optimized for workstation deployment via vLLM and Docker. NVIDIA has already integrated support across its stack, including local execution on RTX GPUs and Jetson Orin Nano for edge AI applications. <a href="https://app.daily.dev/feed-by-ids?id=DxUdrK5VU&amp;id=FucIhSrk0&amp;id=hi4RmWdI4&amp;id=GNDRJXg4G&amp;id=hUcJhitIB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>JavaScript 2026 roadmap features Temporal API and native TypeScript in Node</strong>: The ecosystem is shifting toward native resource management and Go-based compilers for TypeScript v7. <a href="https://app.daily.dev/posts/yHb6Dn8XT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel gains cache aware scheduling for Xeon and EPYC</strong>: New Intel patches colocate tasks sharing the same last-level cache to significantly reduce latency on multi-cache CPUs. <a href="https://app.daily.dev/feed-by-ids?id=3Go8jplR2&amp;id=LLg2zQ9sz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Jujutsu VCS v0.40.0 adds advanced diff matching functions</strong>: The Git-compatible tool now supports <code>diff_lines_added()</code> revsets and TUI improvements for parent/child visualization. <a href="https://app.daily.dev/posts/XNj5BzBCW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vercel Sandbox restores now hit sub-second speeds</strong>: Parallelized S3 streaming and NVMe-backed LRU caching reduced p75 restore times from 40 seconds to under 1 second. <a href="https://app.daily.dev/posts/1Xib7hCkG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Clerk Core 3 bumps free tier to 50K users</strong>: The update introduces a unified Show component and unlimited applications for all plans, including the free tier. <a href="https://app.daily.dev/posts/B73gISWRG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pretext library enables DOM-free text height calculation</strong>: Built by a React core dev, it uses off-screen canvas measurement for high-performance dynamic layouts. <a href="https://app.daily.dev/posts/rS0vBup8W" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Uber first CTO details the platform and program split</strong>: Thuan Pham explains how color-coded sticky notes and Senior 1/2 levels managed hypergrowth before microservices. <a href="https://app.daily.dev/posts/Mz12nMS5l" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The junior death spiral and AI ROI myths</strong>: While executives use ‘AI efficiency’ to justify a 20% drop in entry-level hiring, 94% of companies have yet to see any measurable ROI from generative AI. We’re effectively breaking the talent pipeline to generate code that DORA metrics show is four times more expensive to maintain and hasn’t actually improved delivery speed. <a href="https://app.daily.dev/feed-by-ids?id=KVQvWvGP4&amp;id=jwuM4zMBw&amp;id=uIab7G9MK&amp;id=Qfe9o2iyU&amp;id=zFChurAA9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA marketing slaves and the upscaling crutch</strong>: Hardware reviewers are being called out for acting as ‘marketing slaves’ by prioritizing proprietary DLSS benchmarks over raw rasterization. Developers are now using AI upscaling as a lazy substitute for actual optimization, leading to a decline in PC gaming performance that even high-end hardware can’t brute-force anymore. <a href="https://app.daily.dev/feed-by-ids?id=M5vdbYsqM&amp;id=Vs0fd08bp&amp;id=TpBJpGg67" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Startup methodologies have learned absolutely nothing</strong>: Despite 25 years of Lean Startup and Design Thinking, startup survival rates haven’t budged an inch according to BLS data. These frameworks have become self-defeating ‘cargo cults’ that force founders to converge on identical, uncompetitive strategies while pretending it’s a science. <a href="https://app.daily.dev/posts/N6roJjjoX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
