<h2>Must know</h2>
<ul>
<li><strong>Kubernetes networking gets multi-network multicluster support</strong>: Istio 1.29 is rolling out beta support for ambient multi-network multicluster, which means better telemetry and connectivity across different networks. This update addresses issues where peer metadata was missing in cross-network deployments, making it easier to track traffic. It also allows ingress gateways and waypoint proxies to route requests directly to remote clusters, improving overall network efficiency and observability for complex Kubernetes setups. <a href="https://app.daily.dev/posts/chnxpN00d" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical vulnerabilities found in popular VS Code extensions</strong>: OX Security disclosed critical vulnerabilities in four widely used VS Code extensions, including Live Server and Code Runner, affecting over 128 million installations. These flaws could allow file theft, remote code execution, and local network reconnaissance. While Microsoft silently patched its Live Preview extension, most maintainers haven’t responded, leaving many developers exposed. This highlights a significant supply chain risk within the developer tooling ecosystem. <a href="https://app.daily.dev/posts/IuDICegHN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Copilot chat exposed confidential emails</strong>: Microsoft confirmed a security bug in Copilot Chat that allowed the AI to bypass confidentiality labels and Data Loss Prevention (DLP) policies. This flaw enabled Copilot to summarize emails from users’ Sent Items and Drafts folders without authorization, active since late January. While a fix is rolling out, the incident underscores the ongoing enterprise anxiety about AI tools accessing sensitive corporate data, especially in regulated environments. <a href="https://app.daily.dev/posts/jSmIdhAQU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>PostgreSQL extension for ClickHouse analytics is fast</strong>: The <code>pg_clickhouse</code> extension for PostgreSQL offloads analytical queries to ClickHouse, achieving near-native ClickHouse performance and outperforming other Postgres analytics extensions on ClickBench benchmarks. <a href="https://app.daily.dev/posts/KSlwK5lud" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenFGA improves authorization list objects algorithm</strong>: OpenFGA’s <code>ListObjects</code> algorithm, which enumerates all objects a user has a relation to, has been significantly improved with a concurrent worker pipeline and better handling of complex authorization graphs. <a href="https://app.daily.dev/posts/0Q9jrsmOE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kyverno 1.17 promotes CEL-based policy engine to GA</strong>: Kyverno 1.17 has promoted its Common Expression Language (CEL)-based policy engine to General Availability, offering stable policy types and new functions for Kubernetes policy enforcement, with legacy JMESPath policies deprecated. <a href="https://app.daily.dev/posts/fnlW905uT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches custom agents for AI workflows</strong>: Databricks released Custom Agents, formerly Agent Framework, enabling developers to build, test, and deploy production-quality AI agents as fully managed Databricks Apps with built-in memory and enterprise data access. <a href="https://app.daily.dev/posts/QJn2roYcs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>KEDA scales Kafka sinks with consumer lag</strong>: Trivago successfully used Kubernetes Event-Driven Autoscaling (KEDA) with Kafka scalers to scale Kafka sink services down to zero replicas based on consumer group lag, drastically reducing idle resource consumption. <a href="https://app.daily.dev/posts/vltl2g6Wl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tailscale integrates with Huntress and Fleet for posture enforcement</strong>: Tailscale now integrates with Huntress EDR and Fleet MDM, allowing device posture and management state to be used in Tailscale access policies, automating access restrictions for non-compliant devices. <a href="https://app.daily.dev/feed-by-ids?id=L7iJSIIDS&amp;id=VukD3HNtV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New developer tools for cloud deployment and streaming</strong>: Deno Deploy introduced <code>deno sandbox</code> for instant Linux microVM provisioning, TinyIce offers a lightweight Icecast2-compatible audio streaming server, and Konta brings GitOps workflows to Docker Compose on VPS environments. <a href="https://app.daily.dev/feed-by-ids?id=7Ar3LQ3uM&amp;id=BHwbWnzAf&amp;id=QRFLXVuPF" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s sonnet 4.6 is a token muncher</strong>: Anthropic is out here marketing Claude Sonnet 4.6 as ‘40% cheaper’ with improved benchmarks. But independent analysis from Artificial Analysis says it uses over four times more tokens than its predecessor due to ‘adaptive thinking.’ So, that ‘cheaper’ model might just be munching through your token budget faster than you can say ‘hallucination.’ Classic AI marketing: give with one hand, take with three. <a href="https://app.daily.dev/feed-by-ids?id=i4O6sQBIw&amp;id=j12nF8mSg&amp;id=jj2Kauaop&amp;id=lUz1VvbAr&amp;id=y0N7mzotz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI productivity gains are a myth, says NBER survey</strong>: A new NBER survey of nearly 6,000 executives found that over 80% reported no measurable impact from AI on productivity or employment. Ninety percent saw no gains in the last three years. Yet, these same execs anticipate AI will eliminate 1.75 million jobs by 2028. So, AI isn’t helping now, but it’s definitely coming for your job. Makes perfect sense, right? <a href="https://app.daily.dev/posts/rfmYnNc4e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code is making open source maintainers miserable</strong>: Open-source maintainers, like Godot’s Rémi Verschelde, are drowning in ‘draining and demoralizing’ AI-generated pull requests. These LLM-generated PRs are often nonsensical, verbose, and waste reviewer time. GitHub even acknowledged the problem. So, while AI is supposedly making coding easier, it’s just offloading the garbage collection onto already overworked human maintainers. Progress! <a href="https://app.daily.dev/feed-by-ids?id=HouoEDJq7&amp;id=YaGme2ewn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
