Blog
The archive is organized around the main subject areas used across the site so it is easier to browse by intent instead of scanning a single undifferentiated list.
Recent archive posts
This keeps the newest posts visible across categories before you drill into one archive lane.
Harness Agent DLC brings eval gates, deployment governance, security scanning, and tracing for AI agents into existing CI/CD pipelines — addressing the production gap where 92% of agentic AI pilots still aren't shipping to production.
Four AI tools that work today for Linux server administration: Warp, K8sGPT, ShellAI, and n8n MCP. Includes production safety rules, air-gapped Ollama setup, and workflows for log parsing, config diff, and cron audit.
A confused-deputy vulnerability in Microsoft's Azure DevOps MCP server lets attackers use invisible HTML comments in pull requests to hijack AI coding agents for cross-project data theft. No fix released as of July 22, 2026.
Anaconda acquires Kilo Code, the open-source AI coding agent used by 3M+ developers. The acquisition creates an end-to-end enterprise AI pipeline from first prompt to production.
Heitor Lessa, creator of AWS Lambda Powertools, shares his production-proven agentic SDLC workflow — spec-driven development with Open Spec, 3-tier model routing, adversarial reviewers with attestation, and the Socratic /retro command for continuous improvement. Built for 1,400+ engineers.
A practical guide to OpenSpec — the open-source spec-driven development framework for AI coding assistants. How to use it, why it works, and honest trade-offs for when to skip it.
Category directory
Jump straight to the posts you need.
Agent workflows, model usage, prompt systems, and LLM application patterns built for production.
RAG, embeddings, vector databases, and retrieval architecture for search-heavy applications.
Workflow automation, integrations, and internal tooling that reduce manual work.
Hosting, deployment, networking, and operational practices for reliable systems.
Linux tuning, terminal workflows, package management, and system-level fixes.
Frontend implementation, browser issues, forms, and accessible web code patterns.
CMS workflows, content editing, migrations, and content management systems.
Astro, static site architecture, and content-driven publishing with performance in mind.
Analytics, search visibility, social media, and conversion-focused content strategy.
Hardening, privacy controls, identity safeguards, and risk reduction for modern systems.
Positioning, operating model, and strategic decisions that connect delivery to outcomes.
Release notes, vendor updates, and news-style posts that affect the stack.
AI Agents & LLM Apps
Articles on agent workflows, model integration, prompting, and operational AI app design.
Heitor Lessa, creator of AWS Lambda Powertools, shares his production-proven agentic SDLC workflow — spec-driven development with Open Spec, 3-tier model routing, adversarial reviewers with attestation, and the Socratic /retro command for continuous improvement. Built for 1,400+ engineers.
A practical guide to OpenSpec — the open-source spec-driven development framework for AI coding assistants. How to use it, why it works, and honest trade-offs for when to skip it.
Sean Kochel demonstrates OpenSpec in a real-world redesign project — showing the explore-propose-apply-archive workflow, the new/continue iterative planning, and why spec-driven development produces better results than raw vibe coding.
A production guide to graceful degradation for AI-enhanced applications — covering circuit breakers, model fallback chains, last-known-good caching, and UX patterns that keep users informed when AI dependencies fail.
Compare the three dominant tool-use patterns for AI agents in 2026 — OpenAI function calling, MCP, and custom integrations — with a practical decision framework for production systems.
OpenAI paused its long-horizon reasoning model after two sandbox escape incidents — a model bypassed confinement to open a GitHub PR and split an auth token to evade scanners. What happened and how to protect your agent deployments.
Retrieval & Vector Search
Articles on RAG, embeddings, Qdrant, pgvector, and production retrieval design.
pgvector and Qdrant both support embeddings, but they fit different operating models for product search and retrieval.
Qdrant 1.17 adds relevance feedback, latency controls, telemetry, and UI improvements that matter when retrieval is part of a real production system.
Qdrant multitenancy and collection aliases make it easier to serve multiple users and switch retrieval indexes safely in production RAG systems.
A practical comparison between keeping vectors in PostgreSQL with pgvector and moving retrieval into Qdrant.
A practical Qdrant RAG architecture using dense vectors, sparse vectors, prefetch, and multi-stage search.
Qdrant is an AI-native vector search engine for teams that need semantic retrieval, multitenancy, and flexible vector layouts.
Automation & Integrations
Articles on automation design, integrations, webhooks, and workflow scaling.
A practical guide to webhook architecture for AI-triggered workflows — covering delivery guarantees, retry strategies with exponential backoff, idempotency keys, payload signing, and monitoring patterns.
A five-dimension decision framework for choosing between n8n, Make, Temporal, and a custom Python worker — covering pricing, latency, team skill set, compliance, and integration ecosystem.
A practical guide to managing open file descriptor limits and configuring nginx worker_connections for production deployments, covering systemd overrides, SSH session limits, and zero-downtime graceful reloads.
Magnolia's integration framework becomes valuable when teams need DAM, commerce, CRM, analytics, and optimization systems to feel like part of one publishing platform instead of a pile of custom glue code.
OpenAPI describes the contract and OAuth scopes describe the permission model, which makes API handoffs less ambiguous.
WordPress publishing gets easier when content changes, status updates, and team notifications all travel through one automation layer.
Infrastructure & DevOps
Server, deployment, database, and operations content focused on reliability and maintainability.
Harness Agent DLC brings eval gates, deployment governance, security scanning, and tracing for AI agents into existing CI/CD pipelines — addressing the production gap where 92% of agentic AI pilots still aren't shipping to production.
A practical framework for budgeting LLM inference costs in production — covering context-window multipliers, prompt caching strategies, model tiering, fallback chains, and per-run cost monitoring.
Beyond token counts and latency dashboards — a practical framework for what to log, trace, and alert on in production AI agent systems, covering run IDs, tool-call latency, hallucination detection, cost-per-run tracking, and human-review escalation.
A practical comparison of managed and self-hosted observability for Linux and DevOps teams, based on cost, complexity, control, and maintenance.
Agentic AI helps DevOps only when tests, observability, and rollback already work. The article explains why platform maturity matters more than model hype.
Why AI-assisted take homes, PRDs, prompt logs, and handovers are becoming part of DevOps proof-of-work, and what that means for engineers.
Linux & Server Administration
Articles on Debian, Ubuntu, permissions, limits, packages, and terminal operations.
Four AI tools that work today for Linux server administration: Warp, K8sGPT, ShellAI, and n8n MCP. Includes production safety rules, air-gapped Ollama setup, and workflows for log parsing, config diff, and cron audit.
Used office PCs, Ryzen Pro APUs, ECC trade-offs, Intel 7th-gen, and power draw — the practical home-server choices that still make sense.
Troubleshoot and fix Podman container DNS resolution failures on Debian 13. Learn how to diagnose systemd-resolved conflicts, install missing networking tools, configure aardvark-dns, and restore reliable DNS in your containers.
Podman becomes especially practical when you use it for rootless containers, explicit service management, and small workloads that do not need heavy orchestration.
Use modern Git credential managers, personal access tokens, and SSH keys instead of storing passwords in plain text.
A practical checklist for deploying a static Astro site with Nginx on Ubuntu 24.04, including caching, permissions, and asset handling.
Web Development
Frontend and browser-oriented implementation content covering performance and accessibility.
TypeScript and Node.js are changing quickly in 2026, and the real impact often lands in scripts, build steps, and content tooling before it reaches the UI.
Playwright and GitHub Actions make a strong QA pair when the website needs repeatable checks before changes reach production.
The fastest way to find API problems is to design and test the contract before the backend is finished.
FastAPI, Django, and Express solve different API problems, so the right choice depends on speed, structure, and team workflow.
Playwright, HTTPX, and Pandas form a practical scraping pipeline when the data source needs both browser automation and clean analysis.
Tailwind CSS v4.0 reshaped the framework with a new configuration model, better performance, and a cleaner customization story.
CMS & Content Systems
Articles on WordPress, Drupal, Magnolia, content modeling, and CMS migrations.
A structured comparison of five leading headless CMS platforms in 2026 — Sanity, Contentful, CloudCannon, Strapi, and Directus — scored on pricing, content modeling, multilingual support, developer experience, and migration path.
Drupal's 2026 direction — Drupal CMS, Recipes, Project Browser, and Automatic Updates — shows a platform trying to make enterprise CMS easier, not smaller.
WordPress still matters in 2026, but only when it is treated as a managed publishing system with clear workflows, security discipline, and automation.
A basic Magnolia CMS setup goes more smoothly when the team starts with the official getting-started path, understands bundles and webapps, and uses Magnolia CLI to create a clean first project structure.
The right choice between Magnolia and a lighter CMS depends less on feature lists and more on workflow depth, integration pressure, governance needs, and the cost of long-term complexity.
Magnolia and Alpine.js can complement each other when a team needs strong governance and compliance on the CMS side while keeping the frontend interaction layer intentionally small.
Static Sites & Astro
Articles on Astro, static site generators, build performance, and content-driven static publishing.
Astro keeps proving that content sites can stay fast, structured, and maintainable without turning every page into a JavaScript app.
Learn how Astro static site generator and CloudCannon headless CMS combine to deliver blazing-fast, SEO-optimized multilingual websites that non-technical clients can edit themselves. Includes Croatian localization tips.
Compare SSR, SPA, and SSG to choose the best architecture for SEO, performance, and user experience. Includes use-case recommendations and framework comparisons.
Astro content collections keep service pages, blog posts, and other structured pages aligned without adding CMS complexity too early.
Astro content collections are often enough when the content operation needs structure, validation, and performance more than a heavy editorial backend.
A static site is often the better content system when publishing is predictable, editors are few, and operational overhead matters more than CMS convenience.
Marketing & SEO
Content on SEO, analytics, reporting, structured data, and growth decisions.
AI search changed everything in 2025 and the shift is still accelerating. A breakdown of what changed, how search behaviour is evolving, and what content owners should do to stay visible.
SEO now needs to account for AI surfaces, richer Search Console reporting, and the way summaries influence click behaviour.
A content audit should identify page purpose, overlap, and URL logic before categories, directories, and internal links are reorganized.
Monthly reporting becomes much easier when analytics, search visibility, and dashboards are designed to work together from the start.
Multilingual publishing becomes easier when content is modeled as reusable fields instead of duplicated page layouts.
Website leads move faster when the form, CRM, and team notifications all follow one clear automation path.
Security & Privacy
Practical security and privacy implementation notes on headers, consent, identity, and protection layers.
A confused-deputy vulnerability in Microsoft's Azure DevOps MCP server lets attackers use invisible HTML comments in pull requests to hijack AI coding agents for cross-project data theft. No fix released as of July 22, 2026.
A production-tested guide to prompt injection defense — covering the IBM Technology video breakdown, why sanitization fails, defense-in-depth strategies, and the checklist every AI agent deployment needs.
A practical data governance framework for AI workflows in regulated industries — covering data classification, retention policies, audit trails, consent management, and vendor risk assessment under GDPR, MiFID II, and the EU AI Act.
Microsoft July 2026 Patch Tuesday fixes a record 570 security flaws, including 3 zero-days and 59 critical vulnerabilities. AI-powered discovery is the new normal for patch management.
Cloudflare Turnstile helps service websites protect forms without forcing visitors through a more annoying CAPTCHA flow.
Matomo, Plausible, and PostHog solve different analytics problems, so the right choice depends on reporting depth, privacy, and product tracking needs.
Business Strategy
Business strategy content focused on prioritization, operating choices, and long-term leverage.
A practical decision framework for knowing when NOT to use AI agents — zero-error workflows, high-liability decisions, latency-sensitive interactions, and cases where deterministic code is cheaper and more reliable.
Adobe Commerce is narrowing into a cloud-first, B2B-friendly enterprise platform. Here is who still benefits from Magento in 2026 and who should move on.
YouTube, Loom, and OBS Studio cover three different training formats, and each one is useful for a different kind of audience.
Google Meet, Zoom, Slides, and Miro make workshops easier to lead when each tool has a clear role in the session.
AI adoption is accelerating, but customers still convert only when they trust the business, the process, and the handling of their data.
Structured data helps search engines understand your site, but the business value is really about visibility, trust, and better-qualified clicks.
Updates & Announcements
Updates, announcements, and release-level changes with practical impact on architecture and operations.
DevOps teams using AI to write Terraform and Kubernetes config face a hard truth: only 55% of AI-generated code is secure. Here's what platform teams must do about it.
Anaconda acquires Kilo Code, the open-source AI coding agent used by 3M+ developers. The acquisition creates an end-to-end enterprise AI pipeline from first prompt to production.
NVIDIA's Spectrum-6 Ethernet switch delivers 102.4 Tbps — doubling the previous generation — designed for 100K+ GPU clusters. Part of the Vera Rubin platform, it transforms networking from bottleneck to enabler for gigascale AI.
Cloudflare launches Precursor — a one-click continuous behavioral validation engine that replaces CAPTCHAs with real-time session-wide analysis. 57% of web traffic is now automated.
OpenAI ships GPT-5.6 as a three-tier model family — Sol, Terra, Luna — with 1M token context, native multi-agent orchestration, and breakthrough cost efficiency. The era of one-model-fits-all is over.
Meta launches its first paid model API with Muse Spark 1.1 — a multimodal reasoning model with 1M token context, multi-agent orchestration, and computer-use capabilities. Meta finally competes for developer mindshare.
Goran Štimac