<h2>Must know</h2>
<ul>
<li><strong>European Commission cloud hit by supply chain attack</strong>: The European Commission’s infrastructure was compromised via a supply chain attack targeting the Trivy open-source security scanner, according to EU-CERT and reports from The Next Web. Approximately 92 GB of data, including staff emails and personal info across dozens of EU institutions, was leaked. This follows a similar compromise of the LiteLLM system, highlighting a systemic vulnerability in common security tooling. <a href="https://app.daily.dev/posts/mGRwKz2VN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cursor shifts the IDE from editor to agent console</strong>: The new Cursor 3 release, codenamed Glass, attempts to demote the traditional code editor to a secondary role in favor of a centralized agent management console. As reported by The New Stack, the update introduces multi-repo workspaces and ‘Cloud Handoff’ for session portability, signaling a pivot toward developers acting as system operators who orchestrate AI agents rather than writing lines of code. This move directly counters Anthropic’s terminal-first approach with Claude Code. <a href="https://app.daily.dev/posts/GsQ26ueQa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Karpathy proposes the LLM maintained personal wiki</strong>: Andrej Karpathy has outlined a new pattern for personal knowledge management that moves beyond traditional RAG by using LLMs to incrementally build and maintain a structured markdown wiki. Instead of re-deriving answers from raw docs every time, the LLM integrates new information, flags contradictions, and maintains cross-references within a tool like Obsidian. This approach aims to eliminate the manual maintenance burden that typically causes human-led wikis to fail over time. <a href="https://app.daily.dev/posts/7o1BSKg72" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Neovim 0.12.0 adds native autocomplete and plugin management</strong>: The release introduces ‘vim.pack’ for built-in plugin management and native insert-mode completion to reduce dependency on third-party bloat. <a href="https://app.daily.dev/feed-by-ids?id=3fbVLSQI2&amp;id=T9nUhWoBa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mise and rv are winning the Ruby version manager wars</strong>: The Rust-based ‘mise’ is now the standard for speed, while the new ‘rv’ tool mimics Python’s ‘uv’ by using precompiled binaries for near-instant installs. <a href="https://app.daily.dev/feed-by-ids?id=II5rHR9XP&amp;id=iRc7VAJo1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ryx ORM claims to solve FastAPI database bottlenecks</strong>: A new Python ORM with a Rust core releases the GIL during execution, reportedly reducing database wait times from 87% to 11% in async workloads. <a href="https://app.daily.dev/posts/0TeARnMMJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Usage aware caching tackles AI scraper traffic</strong>: A proposed three-tier caching strategy aims to protect CMS sites from the flat traffic distribution caused by aggressive AI data scraping. <a href="https://app.daily.dev/posts/XwdCsHTFt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tail call optimization comes to nightly Rust</strong>: Using the new ‘become’ keyword, developers are achieving interpreter performance that beats hand-written assembly on ARM64. <a href="https://app.daily.dev/posts/v5awwh224" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TurboQuant brings SIMD vector compression to the browser</strong>: An experimental WASM package enables 6x vector compression with fast dot products for similarity searches in modern browsers. <a href="https://app.daily.dev/posts/h7ZO3PZhH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tabularis integrates SQL notebooks into the DB client</strong>: A new tool aims to kill the context switch between database clients and Jupyter by offering cell-based SQL with CTE-resolved references. <a href="https://app.daily.dev/posts/GJkLoOlRe" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The slow LLM for deep engagement</strong>: A writer is unironically praising a Chrome extension that artificially slows down ChatGPT responses to force ‘active reading.’ Apparently, we’ve reached a point where the solution to AI-induced brain rot is to simulate the speed of a 56k modem. If you need a loading bar to help you think, the problem isn’t the AI’s speed; it’s your attention span. <a href="https://app.daily.dev/posts/A82yOVgkm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vibe coding your way to a total rewrite</strong>: A developer bragged about building a SQLite toolset using ‘vibe-coding’ with Claude, only to admit it produced such a mess of spaghetti code that they had to rewrite the entire thing in Rust. It turns out that ‘vibes’ are a poor substitute for architectural design and basic engineering taste. We’re successfully automating the process of creating technical debt at record speeds. <a href="https://app.daily.dev/posts/fxPxl1DNh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The prompt box as the new enterprise homepage</strong>: Product managers are currently obsessed with replacing structured, functional UIs with a single blank AI chat box, effectively offloading all cognitive load onto the user. As critics point out, removing affordances and muscle memory in favor of ‘exploratory queries’ is just lazy design masquerading as innovation. Not every internal tool needs to be a chatbot that requires a paragraph of text to find a payroll link. <a href="https://app.daily.dev/posts/4kvvQJknf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
