<h2>Must know</h2>
<ul>
<li><strong>Malicious LiteLLM versions on PyPI steal cloud credentials</strong>: PyPI issued an urgent warning regarding two malicious versions of LiteLLM (1.82.7 and 1.82.8) that were live for two hours on March 25. These packages, part of a campaign by TeamPCP, exfiltrated environment variables, SSH keys, and cloud credentials for AWS, GCP, and Azure using a three-stage AES-encrypted payload. Given LiteLLM’s massive footprint in 36% of cloud environments, Wiz and PyPI researchers advise immediate rotation of all secrets if these versions were installed. <a href="https://app.daily.dev/posts/kjJd6k0JC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes SIG Apps debuts Agent Sandbox for AI state</strong>: The Kubernetes SIG Apps group has launched the Agent Sandbox project to address the friction of running stateful AI agents on standard primitives. This new Custom Resource Definition provides a declarative API for single-entity workloads, offering dedicated hostnames and a SandboxWarmPool to eliminate the cold starts that plague agentic workflows. The project supports secure runtimes like gVisor and Kata Containers, signaling a shift toward making Kubernetes a native home for autonomous agents. <a href="https://app.daily.dev/posts/wroMbko1c" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Arm enters the data center silicon race with AGI CPU</strong>: Arm announced its first production silicon, the ARM AGI CPU, specifically architected for AI infrastructure and agentic workloads in modern data centers. Built on 3nm lithography with up to 136 Neoverse V3 cores, the chip supports PCIe Gen6 and CXL 3.0 to handle the massive I/O demands of LLM clusters. Reference designs from Supermicro suggest these chips can scale to over 45,000 cores per liquid-cooled rack, directly challenging incumbent x86 and custom hyperscaler silicon. <a href="https://app.daily.dev/posts/S5lBW0zAU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AWS Load Balancer Controller adds type safe Gateway API support</strong>: Version 3.1.0 replaces brittle annotations with CRDs for NLB and ALB configuration, enabling cross-namespace routing and role-separated RBAC. <a href="https://app.daily.dev/posts/ELZeYfvNA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open source agents for dbt and penetration testing</strong>: Altimate-code provides a data engineering harness for SQL warehouses, while Cyberstrike automates web app exploitation using a unified LLM interface. <a href="https://app.daily.dev/feed-by-ids?id=6Au97kvTm&amp;id=bgY3HXRTU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GrapheneOS rejects age verification and may exit major markets</strong>: The privacy-focused Android fork stated it will not comply with laws in the UK, Australia, and Brazil that mandate data collection at the OS level. <a href="https://app.daily.dev/posts/agWvIPypH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vercel launches elastic build machines to optimize CI costs</strong>: A new beta feature automatically evaluates project complexity to assign Turbo, Enhanced, or Standard build resources without manual provisioning. <a href="https://app.daily.dev/posts/2w3G0XlDk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>KubeVirt v1.8 adds Intel TDX and PCIe NUMA awareness</strong>: The release aligns with Kubernetes 1.35 and introduces confidential VM support alongside performance tuning for AI and HPC workloads. <a href="https://app.daily.dev/posts/2u9pTr13X" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Syncfusion Agent Skills brings component knowledge to LLMs</strong>: A new CLI tool installs playbook files into projects to help AI assistants correctly configure UI libraries and inject required modules. <a href="https://app.daily.dev/posts/3MBz09FW1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MS DOS 2.0 ported to Apple IIe via x86 coprocessor</strong>: A new bridge program allows the 6502 and 8088 CPUs to communicate, enabling most legacy PC software to run on vintage Apple hardware. <a href="https://app.daily.dev/posts/9mxresSa3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The tokenmaxxing productivity trap</strong>: Engineers at Meta and OpenAI are reportedly competing on internal leaderboards for raw token consumption, a trend dubbed ‘tokenmaxxing.’ It is a textbook case of Goodhart’s Law where a metric becomes a target and ceases to be a useful measure, likely resulting in bloated context windows and expensive, recursive agent loops that ship nothing. As only 31% of UK firms see AI ROI, bragging about your monthly token spend is just a high-tech way of admitting you’re burning VC cash for vibes. <a href="https://app.daily.dev/posts/dl9dAiLtW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI data center capacity is mostly vaporware</strong>: Despite the breathless headlines about the AI buildout, an investigation by Where’s Your Ed At reveals that only 33% of announced US data center capacity is actually under development. NVIDIA is effectively selling GPUs for facilities that won’t have power for years, creating a massive gap between ‘sales’ and actual operational utility. The industry is currently a capital misallocation bubble propped up by misleading reporting and a desperate need to justify hyperscaler stock prices. <a href="https://app.daily.dev/posts/X3my4Sf8s" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Jevons Paradox won’t save your developer job</strong>: Optimists are citing the Jevons Paradox to argue that AI efficiency will simply create ‘infinite demand’ for software, keeping every developer employed forever. This ignores the reality that businesses don’t have infinite budgets for maintenance, security, and integration, even if the initial code generation is ‘free.’ It’s a comforting bedtime story for senior management that hand-waves away the structural disruption of the workforce with 19th-century coal economics. <a href="https://app.daily.dev/posts/1powuaBgn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
