The second major Dolt release turns on automatic garbage collection and archival storage by default, cutting database footprints and speeding up queries. Benchmarks show Dolt 2.0 outpacing MySQL on sysbench workloads, while remaining backward‑compatible with 1.x databases.
Moving code review upstream, reviewing design intent and specifications before any code is written, catches defects earlier, cuts CI/CD churn, and shifts the approval gate from merge time to intent capture. With AI-generated code, this upstream focus becomes essential to maintain quality while preserving development speed.
I crawled the 100 k most‑starred GitHub repos and found 73 % contain a detectable task runner. GitHub Actions now accounts for the bulk of CI/CD pipelines, while Make remains the go‑to general‑purpose runner, and fresh projects are gravitating toward modern package‑manager tools such as uv and pnpm. This shift tells engineers where to invest automation effort in 2025.
Pulumi 3.179.0 adds Resource Hooks, letting you attach arbitrary callbacks to create, update, and delete operations. Teams can now run scripts, like opening SSH tunnels or pushing metrics, right before or after a resource changes, turning declarative IaC into a programmable workflow.
Grafana Assistant now integrates with more than 30 observability data sources, including AppDynamics, Azure Monitor, Dynatrace, Honeycomb, New Relic, Splunk, and Zabbix, so a single AI prompt can query across all of them. On-call engineers no longer need to flip between tools or manually stitch results, accelerating troubleshooting and reducing cognitive overload during incidents.
The New Stack details the open‑source EKS Node Monitoring Agent that watches GPU health, surfaces node conditions, and triggers Karpenter to replace faulty GPU instances automatically. Running across thousands of clusters, the team distilled six hard‑won lessons about reason‑code stability, jitter, and scaling self‑repair, showing how to eliminate manual node‑drain toil.
GitHub Copilot and similar AI agents are now generating millions of pull requests, each needing a fresh dev/test environment that lives only minutes. Traditional platform engineering, provisioning full stacks per request, can’t scale; the cost and latency explode. Teams must shift to an AI‑native model that serves environments on demand, like traffic, at agent speed.
Cloudflare swapped ClickHouse for TimescaleDB to power the Zero Trust suite’s time‑series analytics, keeping the stack simple by extending PostgreSQL instead of adding another database. The move let them handle billions of rows while retaining familiar tooling, reducing operational overhead and improving reliability for future product growth.
Netflix replaced Tudum.com's Kafka‑based CQRS pipeline with RAW Hollow, an in‑memory database they built. The switch collapsed data propagation latency, letting homepage content render in single‑digit milliseconds and delivering a noticeably faster fan experience.
Webflow built its own AI‑powered detection pipeline that enriches alerts, auto‑prioritizes severity, and auto‑closes clear false positives. The result: a 200 % jump in detections handled by a handful of security engineers instead of a full SOC. It shows how internal AI can multiply a team’s coverage without buying a stack of vendor tools.
A flaw in Supabase’s Model Context Protocol (MCP) lets an attacker embed malicious prompts in support tickets, causing the AI‑driven IDE assistant to run arbitrary SQL with service_role rights. The exploit can dump every table, bypassing row‑level security. Teams using Supabase MCP must tighten input validation or restrict service_role usage.
Embedding policies in code lets platform engineers enforce security, compliance, and resource standards automatically at every stage of development. Developers receive instant feedback and can provision resources within safe, predefined golden paths, cutting manual checks and reducing drift. The result is faster delivery, clearer audit trails, and a stronger security posture.
You’re caught up.
That was all 12 stories in the 19 July issue.