[{"data":1,"prerenderedAt":655},["ShallowReactive",2],{"blog-post-/blogs/scheduled-ai-agents-lessons":3,"blog-list":360},{"id":4,"title":5,"body":6,"description":340,"extension":341,"meta":342,"navigation":353,"ogImage":355,"path":356,"seo":357,"stem":358,"__hash__":359},"content/blogs/28. scheduled-ai-agents-lessons.md","Scheduled AI Agents — What Weeks of Unattended Runs Teach You",{"type":7,"value":8,"toc":325},"minimark",[9,24,29,54,57,61,77,80,84,87,97,100,126,130,133,154,157,161,164,174,180,183,188,195,209,212,216,219,223,234,245,248,252,286,294,298,318],[10,11,12,13,18,19,23],"p",{},"An agent you watch is a tool. An agent that runs at 04:00 while you sleep is a cron job with opinions. I've spent the last few weeks running a handful of the second kind inside Orca, my fork of an open-source agent-workspace app, on top of ",[14,15,17],"a",{"href":16},"/blogs/oh-my-pi","omp",". This isn't a success story with numbers. It's what broke, what held, and what I'd build into every scheduled agent from day one. The thesis is the same as in ",[14,20,22],{"href":21},"/blogs/harness-vs-model","harness vs. model",": the model is rarely the problem. The guardrails around it are.",[25,26,28],"h2",{"id":27},"whats-running","What's Running",[30,31,32,41,47],"ul",{},[33,34,35,36,40],"li",{},"A ",[37,38,39],"strong",{},"daily upstream-sync agent",": my fork tracks an upstream project, and every day a fresh git worktree starts an agent that follows a runbook, merges upstream, and opens a PR.",[33,42,35,43,46],{},[37,44,45],{},"daily research agent"," that searches, filters, and publishes one HTML page. Its only output is that page.",[33,48,49,50,53],{},"A smaller ",[37,51,52],{},"twice-weekly variant"," of the same pattern.",[10,55,56],{},"All three get a fresh workspace per run and a prompt that points at a file in the repo instead of containing the logic.",[25,58,60],{"id":59},"put-the-logic-in-a-runbook-not-in-the-schedule","Put the Logic in a Runbook, Not in the Schedule",[10,62,63,64,68,69,72,73,76],{},"The automation prompt for the sync job is short: read ",[65,66,67],"code",{},"scripts/sync-upstream.runbook.md"," from ",[65,70,71],{},"origin/main",", follow it step by step, \"it is the single source of truth; it wins over anything you assume.\" Then a few hard rules: never push to ",[65,74,75],{},"main",", never merge the PR, open it as the right GitHub account.",[10,78,79],{},"That split matters. The prompt lives in a scheduler UI where nobody reviews changes. The runbook lives in git, gets reviewed, and changes with the code it describes. When the procedure changes, I edit a markdown file in a PR, not a scheduler field.",[25,81,83],{"id":82},"guard-against-duplicate-work","Guard Against Duplicate Work",[10,85,86],{},"Step 3 of the runbook is the guard: if a sync PR is already open, comment on it and stop. No second branch, no second PR, no merge attempt on top of unreviewed work.",[10,88,89,90,92,93,96],{},"It worked exactly as designed, and it is also the failure mode. Sync PR #11 stayed open for days. Each daily run correctly did nothing except comment with how far upstream had moved: 126 commits past the PR's base one day, 251 ahead of ",[65,91,71],{}," by the last run. By then GitHub reported the PR as ",[65,94,95],{},"CONFLICTING",", and a trial merge showed five conflicts, including a database-migration snapshot where the runbook says to take upstream's version and regenerate the fork's own migration.",[10,98,99],{},"Lessons:",[30,101,102,108,120],{},[33,103,104,107],{},[37,105,106],{},"A one-open-PR guard turns a slow human into a growing backlog."," The guard prevents chaos, but it needs a second signal when the PR has been open too long.",[33,109,110,113,114,116,117,119],{},[37,111,112],{},"The cheapest recovery was closing the PR and syncing fresh."," I allowed the agent to do that. The fresh merge aborted with exit code 2 and roughly 140 conflicting files, so the agent stopped without pushing. That's the right outcome: the job's hard rule is \"never push ",[65,115,75],{},"\" because every push to ",[65,118,75],{}," deploys to production.",[33,121,122,125],{},[37,123,124],{},"Falling far behind upstream is its own cost."," A daily sync is cheap when daily; it gets expensive when skipped for two weeks.",[25,127,129],{"id":128},"idempotent-state-beats-clever-memory","Idempotent State Beats Clever Memory",[10,131,132],{},"The research agent runs unattended once per day and tracks what it has already reported in a state file. The rules I'd copy anywhere:",[134,135,136,142,148],"ol",{},[33,137,138,141],{},[37,139,140],{},"Read the state first, write it last."," If the run dies halfway, tomorrow redoes the work rather than skipping it.",[33,143,144,147],{},[37,145,146],{},"Key on a stable identifier",", not on the text of the item.",[33,149,150,153],{},[37,151,152],{},"Never let the agent \"remember\"."," Memory is a file in the repo or the page itself. A fresh worktree has nothing else.",[10,155,156],{},"This is the same idea as an idempotent queue consumer: a rerun must be safe.",[25,158,160],{"id":159},"publishing-is-the-step-that-fails","Publishing Is the Step That Fails",[10,162,163],{},"The research agent's only deliverable is a page published through the Orca CLI. One morning every run failed at the last step with:",[165,166,172],"pre",{"className":167,"code":169,"language":170,"meta":171},[168],"language-text","Error: No value provided for input HTTP label: Bucket.\n","text","",[65,173,169],{"__ignoreMap":171},[10,175,176,177],{},"It failed for the real page and for a tiny test page with the most restrictive visibility. Auth was valid and the host healthy. The agent's own conclusion, labelled as inference: a missing storage bucket in the pages backend, which it can't fix from inside a run. It reported the failure instead of faking a URL. That's correct behavior and the thing I check first in any unattended agent: ",[37,178,179],{},"does a failed final step produce a loud failure, or a plausible lie?",[10,181,182],{},"Cause on my side: the pages feature stores content in object storage behind a Cloudflare Worker, and on the self-hosted instance that storage hadn't been enabled. Enabling R2 needs a payment method on the account even though the free tier covers small use, and the Worker needs wildcard DNS and two matching secrets between Worker and API. That's a deployment checklist, not agent work.",[184,185,187],"h3",{"id":186},"the-free-plan-tls-trap","The free-plan TLS trap",[10,189,190,191,194],{},"While debugging page links I hit ",[65,192,193],{},"ERR_SSL_VERSION_OR_CIPHER_MISMATCH",". The Worker's TLS was fine. The cause was the hostname shape:",[30,196,197,203],{},[33,198,199,202],{},[65,200,201],{},"\u003Cid>-frame.example.com"," — one label under the wildcard, covered by the free universal certificate.",[33,204,205,208],{},[65,206,207],{},"\u003Cid>.frame.example.com"," — two labels, not covered, handshake fails.",[10,210,211],{},"curl against the Cloudflare IP with each name showed it in a minute. The remaining bug was older builds still generating the dotted form, because the API builds page links from shared code. Rule: on Cloudflare's free plan, keep per-item subdomains to one label.",[25,213,215],{"id":214},"notifications-only-on-completion","Notifications: Only on Completion",[10,217,218],{},"My notification rule for agents: sound when a task finishes or needs me, never for housekeeping. In Orca I removed the sound on closing a tab or session and on closing all ports. It sounds trivial. With several agents running, noise that isn't a signal trains you to ignore the sound that is one.",[25,220,222],{"id":221},"tame-the-review-bot-too","Tame the Review Bot Too",[10,224,225,226,229,230,233],{},"The same discipline applies to a review bot. A team's GitHub Copilot review ruleset had ",[65,227,228],{},"review_on_push"," and ",[65,231,232],{},"review_draft_pull_requests"," both on. The complaint: handle comments, mark ready, and the bot finds new points again. Reading the docs, there's no \"only on new commits\" switch. The options were:",[134,235,236,239,242],{},[33,237,238],{},"Turn off draft reviews: drafts stay quiet, one review at ready, then one per push.",[33,240,241],{},"Turn off review on push: one review at ready, re-request manually after big changes.",[33,243,244],{},"Remove the rule and request the review by hand.",[10,246,247],{},"I recommended option 2 because it fits \"bot first, humans second\". I didn't reproduce the draft-to-ready behavior; that's the docs plus inference.",[25,249,251],{"id":250},"where-it-still-breaks","Where It Still Breaks",[30,253,254,260,274,280],{},[33,255,256,259],{},[37,257,258],{},"Silent staleness."," A guard that says \"stop\" produces no failure, only a growing gap.",[33,261,262,265,266,269,270,273],{},[37,263,264],{},"Environment drift."," The mobile client once called localhost because a root ",[65,267,268],{},".env"," was loaded with ",[65,271,272],{},"override: true"," and beat the command-line values.",[33,275,276,279],{},[37,277,278],{},"Tooling flakiness."," A release upload died on a TLS timeout and had to be retried. One MCP check showed three servers connected and one failed.",[33,281,282,285],{},[37,283,284],{},"Slow browser bridges."," I was chasing a browser relay that made requests feel slow. I don't have a finished root cause to share.",[10,287,288,289,293],{},"None of that is model quality. For the review habits that apply here too, see ",[14,290,292],{"href":291},"/blogs/claude-code-freelance-workflows","Claude Code in real client work",".",[25,295,297],{"id":296},"my-checklist","My Checklist",[134,299,300,303,306,309,312,315],{},[33,301,302],{},"Logic in a versioned runbook, prompt only points to it.",[33,304,305],{},"A guard against duplicate work, plus an alert when the guard fires repeatedly.",[33,307,308],{},"State in a file, read first, written last.",[33,310,311],{},"A final step that fails loudly.",[33,313,314],{},"Hard rules about what the agent may never touch.",[33,316,317],{},"Notifications only for completion or needs-you.",[10,319,320,321,293],{},"If you want agents doing recurring work for your shop or team without surprises, ",[14,322,324],{"href":323},"/contact","Get in touch",{"title":171,"searchDepth":326,"depth":326,"links":327},2,[328,329,330,331,332,336,337,338,339],{"id":27,"depth":326,"text":28},{"id":59,"depth":326,"text":60},{"id":82,"depth":326,"text":83},{"id":128,"depth":326,"text":129},{"id":159,"depth":326,"text":160,"children":333},[334],{"id":186,"depth":335,"text":187},3,{"id":214,"depth":326,"text":215},{"id":221,"depth":326,"text":222},{"id":250,"depth":326,"text":251},{"id":296,"depth":326,"text":297},"Notes from running scheduled, unattended AI agents for a few weeks — a daily upstream-sync job, a research agent that publishes a page, and review-bot tuning. Guards, state files, failing publish steps, and where it breaks.","md",{"seoTitle":343,"date":344,"image":345,"imageWidth":346,"imageHeight":347,"alt":348,"tags":349,"author":352,"published":353,"featured":354},"Scheduled AI Agents: Lessons From Unattended Runs","06.10.2026","/blogs-img/scheduled-agents.webp",1536,1024,"Scheduled agent runs flowing through a guard into a published page, with one failing publish step highlighted",[350,351],"ai","devops","Marco Faul",true,false,"/blogs-img/scheduled-agents.png","/blogs/scheduled-ai-agents-lessons",{"title":5,"description":340},"blogs/28. scheduled-ai-agents-lessons","4PMiHndqEPpzK_K5ttDIIekbUwJ9DOG7dZAXBH6MTgM",[361,372,383,393,403,414,423,431,441,451,461,470,480,489,498,506,515,524,534,544,554,557,568,577,587,598,607,616,625,635,645],{"path":362,"title":363,"description":364,"meta":365},"/blogs/shopware-6-performance-guide","Shopware 6 Developer Performance Guide - Essential Improvements and Fixes","Performance optimization is crucial for any e-commerce platform. Here are the most impactful performance improvements and fixes for Shopware 6 developers.",{"seoTitle":366,"date":367,"image":368,"imageWidth":346,"imageHeight":347,"alt":363,"tags":369,"published":353,"featured":354},"Shopware 6 Performance Guide","24.10.2025","/blogs-img/shopware-performance.webp",[370,371],"shopware","performance",{"path":373,"title":374,"description":375,"meta":376},"/blogs/shopware-frontends-composable-storefront","Shopware Frontends - Building a Composable Storefront with Vue and Nuxt","Shopware Frontends is the official way to build headless storefronts on Shopware 6. Learn the Vue Starter Template, Nuxt layers, composables, and the Store API architecture.",{"seoTitle":377,"date":378,"image":379,"imageWidth":346,"imageHeight":347,"alt":380,"tags":381,"author":352,"published":353,"featured":354},"Shopware Frontends - Composable Storefronts","06.06.2026","/blogs-img/shopware-frontends.webp","Shopware Frontends composable storefront with Vue and Nuxt",[370,382],"frontend",{"path":384,"title":385,"description":386,"meta":387},"/blogs/headless-shopware-cost","What Does a Headless Shopware Store Really Cost? A Realistic 2026 Breakdown","Thinking about going headless with Shopware? An honest breakdown of what a headless storefront actually costs - build, hosting, maintenance - and when it pays off.",{"seoTitle":388,"date":389,"image":390,"imageWidth":346,"imageHeight":347,"alt":391,"tags":392,"author":352,"published":353,"featured":354},"What a Headless Shopware Store Really Costs","02.06.2026","/blogs-img/headless-cost.webp","What a headless Shopware store really costs in 2026",[370,382],{"path":394,"title":395,"description":396,"meta":397},"/blogs/headless-vs-traditional-shopware-2026","Headless vs. Traditional Shopware - How to Choose in 2026","Headless or the classic Twig storefront? A practical 2026 decision framework for Shopware 6 - performance, cost, flexibility, SEO, and team fit - with a clear recommendation.",{"seoTitle":398,"date":399,"image":400,"imageWidth":346,"imageHeight":347,"alt":401,"tags":402,"author":352,"published":353,"featured":354},"Headless vs. Traditional Shopware in 2026","30.05.2026","/blogs-img/headless-vs-traditional.webp","Headless vs traditional Shopware 6 decision framework 2026",[370,382],{"path":404,"title":405,"description":406,"meta":407},"/blogs/supply-chain-security-2026","The Supply Chain Is the New Attack Surface - npm Worms and Slopsquatting in 2026","A self-replicating npm worm and AI-hallucinated packages are the supply-chain threats hitting JavaScript developers right now. Here's how they work and how I lock my projects down.",{"seoTitle":408,"date":409,"image":410,"imageWidth":346,"imageHeight":347,"alt":411,"tags":412,"author":352,"published":353,"featured":353},"npm Worms and Slopsquatting - Supply Chain 2026","11.06.2026","/blogs-img/supply-chain-security.webp","npm supply chain worms and slopsquatting - 2026 developer security threats",[413,351],"security",{"path":21,"title":415,"description":416,"meta":417},"The Real Variable in AI Coding Isn't the Model — It's the Harness","Benchmarks show that switching edit formats improved Grok Code Fast by 10x — zero training compute required. As a developer who relies on AI agents daily, the harness deserves far more scrutiny than we give it.",{"seoTitle":418,"date":419,"image":420,"imageWidth":346,"imageHeight":347,"alt":421,"tags":422,"author":352,"published":353,"featured":353},"AI Coding - The Harness Beats the Model","12.06.2026","/blogs-img/harness-vs-model.webp","AI model connected to a codebase through a harness bridge",[350,351],{"path":16,"title":424,"description":425,"meta":426},"Oh My Pi — The Open-Source Coding Agent I Actually Use Every Day","Oh My Pi is a fork of Mario Zechner's Pi coding agent with ~1,300 commits of incremental improvements. Here's what makes it different, why harness quality shapes agent reliability more than model choice, and why I run it instead of Claude Code.",{"seoTitle":427,"date":419,"image":428,"imageWidth":346,"imageHeight":347,"alt":429,"tags":430,"author":352,"published":353,"featured":354},"Oh My Pi - My Daily Open-Source Coding Agent","/blogs-img/oh-my-pi.webp","Oh My Pi open-source coding agent harness with flowing code streams",[350,351],{"path":432,"title":433,"description":434,"meta":435},"/blogs/1password-claude-agentic-security","1Password for Claude - AI Agents That Log In Without Ever Seeing Your Password","1Password and Anthropic shipped a zero-exposure integration that lets Claude log into websites and finish real tasks without the model ever touching a credential. Here's how it works, what it's good for, and where the limits are.",{"seoTitle":436,"date":437,"image":438,"imageWidth":346,"imageHeight":347,"alt":439,"tags":440,"author":352,"published":353,"featured":354},"1Password for Claude - Logins Without Secrets","20.07.2026","/blogs-img/1password-claude.webp","1Password and Anthropic zero-exposure credential access for Claude AI agents",[413,350],{"path":442,"title":443,"description":444,"meta":445},"/blogs/nuxt-4-5-road-to-nuxt-5","Nuxt 4.5 - Vite 8, Rspack 2, SSR Streaming, and the Road to Nuxt 5","Nuxt 4.5 is the biggest release in a while - Vite 8, an Rsbuild-powered Rspack 2 builder, experimental SSR streaming, stable error codes, and a pile of groundwork for Nuxt 5. Here's what actually matters, what to test before upgrading, and why the Nuxt 3 clock is about to run out.",{"seoTitle":446,"date":447,"image":448,"imageWidth":346,"imageHeight":347,"alt":449,"tags":450,"author":352,"published":353,"featured":354},"Nuxt 4.5 - Vite 8 and the Road to Nuxt 5","18.07.2026","/blogs-img/nuxt-4-5.webp","Nuxt 4.5 release - Vite 8, Rspack 2, SSR streaming and the road to Nuxt 5",[382,351],{"path":452,"title":453,"description":454,"meta":455},"/blogs/netlify-cloudflare-cache-purge","Purging Cloudflare's Cache Automatically on Every Netlify Deploy","This site sits on Netlify behind Cloudflare, and after every deploy Cloudflare kept serving stale HTML. Here's why the Netlify-CDN-Cache-Control header was a dead end for static files, and the 26-line build plugin that purges Cloudflare's edge cache the moment a deploy goes live — including the CJS/ESM gotcha that broke the first build.",{"seoTitle":456,"date":457,"image":458,"imageWidth":346,"imageHeight":347,"alt":459,"tags":460,"author":352,"published":353,"featured":354},"Auto-Purge Cloudflare Cache on Netlify Deploys","22.07.2026","/blogs-img/netlify-cloudflare-purge.webp","Netlify deploy triggering an automatic Cloudflare edge cache purge",[371,351],{"path":462,"title":463,"description":464,"meta":465},"/blogs/lcp-debugging-case-study","Debugging LCP on a Real Site - A Case Study on marcofaul.de","A first-person walkthrough of fixing LCP on this very site - a JS reveal animation gating first paint, a font preload that held rendering hostage, a 5.8s image CDN cold start, and the measurement noise that almost sent me chasing ghosts.",{"seoTitle":466,"date":457,"image":467,"imageWidth":346,"imageHeight":347,"alt":468,"tags":469,"author":352,"published":353,"featured":354},"Debugging LCP - A Real-World Case Study","/blogs-img/lcp-debugging.webp","Debugging Largest Contentful Paint - a real-world case study on marcofaul.de",[371,382],{"path":471,"title":472,"description":473,"meta":474},"/blogs/clean-code-summary-robert-martin","Summary of Clean Code by Robert C. Martin - Essential Guidelines for Better Software","Code is clean if it can be understood easily by everyone on the team. A comprehensive summary of Robert C. Martin's Clean Code principles and best practices.",{"seoTitle":475,"date":476,"image":477,"imageWidth":346,"imageHeight":347,"alt":472,"tags":478,"published":353,"featured":354},"Clean Code by Robert C. Martin - Summary","25.10.2025","/blogs-img/clean-code.webp",[479],"best-practices",{"path":481,"title":482,"description":483,"meta":484},"/blogs/pg-boss-10-to-12-upgrade","Upgrading pg-boss 10 → 12 When There Is No Migration Path","pg-boss 12 cannot migrate a v10 schema - v11 moved job storage to partitioned tables and the migration chain starts after that cliff. Here's how I upgraded a production job queue anyway - drop the schema, rebuild queue state from my own tables, and the job-ownership guard the rollout taught me to add.",{"seoTitle":485,"date":457,"image":486,"imageWidth":346,"imageHeight":347,"alt":487,"tags":488,"author":352,"published":353,"featured":354},"pg-boss 10 to 12 Upgrade Without a Migration Path","/blogs-img/pg-boss-upgrade.webp","Postgres-backed job queue being rebuilt during a pg-boss major version upgrade",[479,351],{"path":490,"title":491,"description":492,"meta":493},"/blogs/building-an-seo-crawler","I Built My Own SEO Crawler - Architecture Notes from seo-pulse","Instead of paying for Screaming Frog or Ahrefs, I built my own SEO audit tool - a Bun monorepo with a Nuxt 4 UI, a pg-boss worker that crawls and scores sites, and an MCP server so AI agents can trigger crawls and read findings. Here's the real architecture, and the parts that hurt - deploys killing workers mid-crawl, queue schema upgrades, and Docker builds dying without an error.",{"seoTitle":494,"date":457,"image":495,"imageWidth":346,"imageHeight":347,"alt":496,"tags":497,"author":352,"published":353,"featured":354},"Building an SEO Crawler - Architecture Notes","/blogs-img/seo-pulse-crawler.webp","Architecture diagram of a self-hosted SEO crawler with web app, Postgres queue, and worker",[350,351],{"path":291,"title":499,"description":500,"meta":501},"Claude Code in Real Client Work — Workflows, MCP, and Where It Breaks","A year of running Claude Code on actual freelance and agency projects — the workflows that hold up, why MCP is the real game changer for client work, the failure modes nobody puts in demos, and what I promise clients about AI-assisted development.",{"seoTitle":502,"date":457,"image":503,"imageWidth":346,"imageHeight":347,"alt":504,"tags":505,"author":352,"published":353,"featured":354},"Claude Code in Freelance Work - Workflows & Limits","/blogs-img/claude-code-workflows.webp","AI coding agent connected to real project tooling in a client development workflow",[350,351],{"path":507,"title":508,"description":509,"meta":510},"/blogs/shopware-mcp-ai-agents","Shopware 6 Meets AI Agents - What an MCP Server for Your Shop Could Do","MCP gives AI agents a standard way to talk to real systems. Here's what an MCP server over Shopware's Admin API could look like - the tool surface, the architecture, and the honest risks of letting an agent near a live shop.",{"seoTitle":511,"date":457,"image":512,"imageWidth":346,"imageHeight":347,"alt":513,"tags":514,"author":352,"published":353,"featured":353},"Shopware 6 + AI Agents - An MCP Server for Your Shop","/blogs-img/shopware-mcp.webp","AI agent connected to a Shopware 6 shop through an MCP server",[370,350],{"path":516,"title":517,"description":518,"meta":519},"/blogs/jtl-to-shopify-migration","Migrating a JTL-Shop to Shopify: What to Import and What to Skip","There is no one-click connector from JTL-Shop to Shopify. Here is how I planned a real migration - what I imported, what I deliberately did not, how scripts verify the result, how to stop notification mails during an order import, and the launch checklist.",{"seoTitle":520,"date":344,"image":521,"imageWidth":346,"imageHeight":347,"alt":522,"tags":523,"author":352,"published":353,"featured":354},"JTL-Shop to Shopify Migration: Plan and Verify","/blogs-img/jtl-shopify-migration.webp","Legacy shop with an encrypted database feeding a verified import pipeline into a Shopify storefront",[350,351],{"path":525,"title":526,"description":527,"meta":528},"/blogs/shopware-configurator-share-links","Shareable Product Configurator in Shopware 6 — State in the URL","How I built a product configurator for a Shopware 6 storefront where the whole state lives in a ?cfg= URL parameter — what to serialize, how the PDF export works, what breaks, and the trade-offs of letting users edit a shared configuration.",{"seoTitle":529,"date":530,"image":531,"imageWidth":346,"imageHeight":347,"alt":532,"tags":533,"author":352,"published":353,"featured":354},"Shopware 6 Configurator: Share Links & PDF Export","02.10.2026","/blogs-img/shopware-configurator.webp","Product configurator card whose state is encoded into a share link that fans out to a phone, a PDF and an email, with an orange loop marking the edit-a-shared-state problem",[370,382],{"path":535,"title":536,"description":537,"meta":538},"/blogs/shopify-personalized-products-and-packaging-fees","Personalized Products and Packaging Fees in Shopify","How I modeled printed cups with tiered prices and a logo upload in Shopify, and the options I weighed for passing a per-item packaging licence fee through to customers — with the pitfalls I hit.",{"seoTitle":539,"date":540,"image":541,"imageWidth":346,"imageHeight":347,"alt":542,"tags":543,"author":352,"published":353,"featured":354},"Shopify Personalized Products & Packaging Fees","04.10.2026","/blogs-img/shopify-personalization.webp","Product configurator with size and quantity tiles and a logo upload, next to a cart that separates a regulatory packaging fee line",[479,351],{"path":545,"title":546,"description":547,"meta":548},"/blogs/proving-where-data-stops-integration-debugging","Proving Where Data Stops — How I Debug Multi-System Integrations","A shop field that never updates, a CDN that can't find its images, a sync that \"randomly\" skips entries. A method for multi-system integrations — instrument every hop, read the boundary evidence, hand over proof instead of opinions — with five real cases from the last weeks.",{"seoTitle":549,"date":550,"image":551,"imageWidth":346,"imageHeight":347,"alt":552,"tags":553,"author":352,"published":353,"featured":354},"Integration Debugging - Prove Where Data Stops","05.10.2026","/blogs-img/integration-debugging.webp","Data pipeline across four connected systems with a check at every hop and one orange marker where the data flow breaks",[479,351],{"path":356,"title":5,"description":340,"meta":555},{"seoTitle":343,"date":344,"image":345,"imageWidth":346,"imageHeight":347,"alt":348,"tags":556,"author":352,"published":353,"featured":354},[350,351],{"path":558,"title":559,"description":560,"meta":561},"/blogs/bfsg-accessibility-online-shops","BFSG for Online Shops — What an Accessibility Audit Actually Finds","The BFSG has applied to B2C online shops since 28 June 2025. What an accessibility audit on a real headless Shopware storefront turned up, which fixes a developer can ship right away, which ones need design sign-off, and a checklist for your own shop.",{"seoTitle":562,"date":563,"image":564,"imageWidth":346,"imageHeight":347,"alt":565,"tags":566,"author":352,"published":353,"featured":354},"BFSG for Online Shops: Real Audit Findings","14.10.2026","/blogs-img/bfsg-accessibility.webp","Online shop wireframe with keyboard focus moving through header buttons, navigation and a product card, each checked against an accessibility list",[567,382],"accessibility",{"path":569,"title":570,"description":571,"meta":572},"/blogs/shopware-decoration-best-practices","Shopware Decoration Best Practices: Override One Method, Keep Others Untouched","Learn the golden rule of Shopware decorations - how to efficiently override only what you need while letting PHP inheritance handle the rest automatically.",{"seoTitle":573,"date":574,"image":575,"imageWidth":346,"imageHeight":347,"alt":570,"tags":576,"author":352,"published":353,"featured":353},"Shopware 6 Decoration Best Practices","26.10.2025","/blogs-img/shopware-decoration.webp",[370,479],{"path":578,"title":579,"description":580,"meta":581},"/blogs/storefront-requests-531-to-101","531 Requests on a Homepage — Cutting a Headless Storefront Down to 101","A headless Shopware 6 + Nuxt storefront loaded 531 requests on its mobile homepage and scored 56 in Lighthouse. After one batch of changes on production it was 101 requests and 88. How I measured, where the weight came from, what I reverted, and a checklist for your own storefront.",{"seoTitle":582,"date":583,"image":584,"imageWidth":346,"imageHeight":347,"alt":585,"tags":586,"author":352,"published":353,"featured":354},"Headless Storefront Performance: 531 to 101 Requests","29.10.2026","/blogs-img/storefront-requests.webp","A storefront waterfall chart shrinking from a dense wall of requests to a short, tidy list, with a performance gauge moving from orange to green",[371,370],{"path":588,"title":589,"description":590,"meta":591},"/blogs/ai-agent-readiness-website","Making Your Website Readable for AI Agents — What Actually Matters","llms.txt, agents.md, ai-catalog.json, AI-bot rules in robots.txt. I shipped all of it on marcofaul.de, then audited it and found four of my own files broken or missing. What the 2026 data says about these files, what really decides whether AI tools can read your site, and a checklist to verify it.",{"seoTitle":592,"date":593,"image":594,"imageWidth":346,"imageHeight":347,"alt":595,"tags":596,"author":352,"published":353,"featured":354},"AI Agent Readiness: llms.txt, AGENTS.md and What Matters","31.10.2026","/blogs-img/agent-readiness.webp","A website feeding server-rendered HTML, structured data and small machine-readable files to AI crawlers, with one broken file highlighted",[350,597],"seo",{"path":599,"title":600,"description":601,"meta":602},"/blogs/headless-gamechanger","Why Headless Shopware 6 with Nuxt is a Game-Changer for Your Online Store","Going headless with Shopware 6 and Nuxt isn't just a tech trend—it's a competitive advantage. Discover why separating your frontend from backend delivers blazing-fast performance, incredible flexibility, and happier customers.",{"seoTitle":603,"date":574,"image":604,"imageWidth":346,"imageHeight":347,"alt":605,"tags":606,"published":353,"featured":353},"Headless Shopware 6 with Nuxt - Why It Wins","/blogs-img/shopware-nuxt-headless.webp","Headless Shopware 6 with Nuxt - The future of e-commerce",[370,382],{"path":608,"title":609,"description":610,"meta":611},"/blogs/quality-tools-that-matter","Code Quality Tools That Actually Matter - Your PR Review Checklist","Stop shipping bugs and messy code. These essential code quality tools catch issues before they hit production, making your team faster and your code cleaner.",{"seoTitle":612,"date":574,"image":613,"imageWidth":346,"imageHeight":347,"alt":614,"tags":615,"published":353},"Code Quality Tools That Actually Matter","/blogs-img/code-quality-tools.webp","Essential code quality tools for modern development",[479,351],{"path":617,"title":618,"description":619,"meta":620},"/blogs/shopware-67-migration-guide","Shopware 6.6 to 6.7 Migration Guide - Breaking Changes and a Safe Upgrade Path","Upgrading Shopware 6.6 to 6.7? Here are the real breaking changes - Vite, Symfony 7.4, Pinia, delayed cache - and a tested upgrade path that won't break production.",{"seoTitle":621,"date":409,"image":622,"imageWidth":346,"imageHeight":347,"alt":623,"tags":624,"author":352,"published":353,"featured":353},"Shopware 6.6 to 6.7 Migration Guide","/blogs-img/shopware-67-migration.webp","Shopware 6.6 to 6.7 migration guide - breaking changes and upgrade path",[370,351],{"path":626,"title":627,"description":628,"meta":629},"/blogs/nuxt-4-migration-guide","Migrating to Nuxt 4 - The New app/ Directory, Smarter Data Fetching, and a Painless Upgrade","Nuxt 4 is here. Learn the new app/ directory structure, improved useFetch/useAsyncData, the path-alias gotcha, and how to upgrade safely with codemods.",{"seoTitle":630,"date":631,"image":632,"imageWidth":346,"imageHeight":347,"alt":633,"tags":634,"author":352,"published":353,"featured":353},"Nuxt 4 Migration Guide","10.06.2026","/blogs-img/nuxt-4-migration.webp","Migrating to Nuxt 4 - new app directory and data fetching",[382,351],{"path":636,"title":637,"description":638,"meta":639},"/blogs/core-web-vitals-inp-2026","Core Web Vitals 2026 - How to Master INP (Interaction to Next Paint)","INP is the Core Web Vital most sites fail. Learn what INP measures, the 200ms threshold, and concrete JavaScript techniques to make your store feel instant.",{"seoTitle":640,"date":641,"image":642,"imageWidth":346,"imageHeight":347,"alt":643,"tags":644,"author":352,"published":353,"featured":354},"Core Web Vitals 2026 - Mastering INP","04.06.2026","/blogs-img/core-web-vitals-inp.webp","Master INP Interaction to Next Paint Core Web Vitals 2026",[371,382],{"path":646,"title":647,"description":648,"meta":649},"/blogs/ai-product-search-shopware","AI-Powered Product Search in Shopware - Semantic and Vector Search That Converts","Keyword search misses what shoppers mean. Here's how to add AI semantic/vector search to Shopware 6 with embeddings - the architecture, the tradeoffs, and a practical build.",{"seoTitle":650,"date":651,"image":652,"imageWidth":346,"imageHeight":347,"alt":653,"tags":654,"author":352,"published":353,"featured":353},"AI Product Search in Shopware","09.06.2026","/blogs-img/ai-product-search.webp","AI-powered semantic and vector product search in Shopware 6",[370,350],1793447960745]