--- title: "Playbooks | StudioMeyer Academy" description: "Schnellstart-Anleitungen im 10-Schritte-Format. Von ChatGPT zu Claude. Erster MCP-Server in 90 Minuten. Memory portabel nutzen. Kein Terminal, kein Tech-Talk." canonical: "https://studiomeyer.academy/en/playbooks" language: en image: "https://studiomeyer.academy/api/og/page/share?title=Playbooks§ion=Playbooks&locale=en" markdown_versions: ["https://studiomeyer.academy/playbooks.md", "https://studiomeyer.academy/en/playbooks.md", "https://studiomeyer.academy/es/playbooks.md"] publisher: "StudioMeyer Academy, https://studiomeyer.academy (llms.txt: https://studiomeyer.academy/llms.txt)" --- Playbooks # Schnellstart-Anleitungen. *Keine Theorie.* Jedes Playbook ist eine konkrete Anleitung. Zehn Schritte, jeder mit einfacher Erklärung und einem Tipp. Kein Terminal, kein Jargon, nichts zum Durchklicken. Die Dinge die Du genau einmal nachlesen willst, bevor Du sie machst. 107 Playbooks · alle kostenlos 10 Schritte pro Playbook 15-90 Min typische Lesezeit DE Sprache · Du-Form, Reddit-Stil Switching over 01 From ChatGPT to Claude, guide in 10 steps No terminal, no tech-talk. Ten steps, each with a simple explanation and a tip. https://studiomeyer.academy/playbooks/von-chatgpt-zu-claude 02 From ChatGPT to Codex, guide in 10 steps Since the July rebuild, Codex sits inside the ChatGPT app and is included in every plan, the free one too. Ten steps from first task to habit, with the limits that actually bite. https://studiomeyer.academy/playbooks/von-chatgpt-zu-codex Career & job 69 AI for the job hunt: 10 steps through the ATS filter How to build a CV with ChatGPT, Claude or Gemini that gets past the automated applicant-tracking software. With the concrete keyword process, real numbers and copy-paste prompts. https://studiomeyer.academy/playbooks/ki-fuer-jobsuche 91 AI as a learning coach: teaching yourself a new field for a career change How to use ChatGPT or Claude to teach yourself a new subject without falling into the usual traps. Ten steps from the first study plan to a project you can show. https://studiomeyer.academy/playbooks/ki-lern-coach-berufswechsel 92 The AI operator job: what do you actually do? An honest look at the profession behind this academy. What an AI operator really does, which skills count, how to get in and who it pays off for. https://studiomeyer.academy/playbooks/berufsbild-ai-operator Setup 04 Using memory portably, same memory in Claude, Codex and Cursor External storage that all your AI tools read from and write to at the same time. One source of truth, many clients. https://studiomeyer.academy/playbooks/memory-portabel-nutzen 08 Hooks against hallucinations Seven concrete Claude Code hooks that stop the model from inventing things, breaking conventions or running destructive commands. https://studiomeyer.academy/playbooks/hooks-gegen-halluzinationen 12 Claude Code update strategy, without your workflow suddenly breaking In April 2026 Anthropic published a postmortem on a quality regression. Here are 10 steps to roll in updates so your workflow doesn't suddenly get worse, with pinning, an eval set and a rollback path. https://studiomeyer.academy/playbooks/claude-code-update-strategie 13 Claude Code cost controls for daily drivers, staying under 30 dollars a day in 10 steps How to control token usage, pick the right model and reduce MCP overhead. Concrete slash commands, settings, patterns from real daily use. https://studiomeyer.academy/playbooks/claude-code-cost-controls-fuer-daily-driver 17 Plan Mode in Claude Code, the invisible seatbelt, when you need it and how to keep it on Plan Mode is the most underrated permission in Claude Code. When it makes sense, how to make Shift+Tab a routine, and where it gets annoying. https://studiomeyer.academy/playbooks/plan-mode-richtig-nutzen 18 Debugging MCP when nothing works, the 10 checks that solve 90 percent of cases Your MCP server doesn't show up, tools are missing, auth breaks. This order saves you the two hours of searching. https://studiomeyer.academy/playbooks/mcp-debuggen-wenn-nichts-geht 18 Mastering prompt caching, from cache miss to an 80 percent hit rate in 10 steps How to set cache\_control properly, build stable prefixes and read cache diagnostics. Concrete patterns for the Claude Code daily driver, your own scripts and MCP servers. https://studiomeyer.academy/playbooks/prompt-caching-meistern 19 Repair memory drift, when your agent's memory starts contradicting itself Diagnosis and repair in 90 minutes when your memory server has gone contradictory, duplicated or stale. https://studiomeyer.academy/playbooks/memory-drift-reparieren 27 Claude Code in a Team, how to keep CLAUDE.md alive without turning it into a trash bin Three devs, one repo, one Claude Code setup. Here are 10 steps to share CLAUDE.md, hooks and slash commands so everyone gets the same output and nobody commits their private notes. https://studiomeyer.academy/playbooks/claude-code-im-team-claude-md-gemeinsam-pflegen 27 Loading third-party Claude Code plugins from ZIP and URL, safely in 20 minutes Since Claude Code 2.1.128 (May 2026) you can load plugins with --plugin-url and --plugin-dir, no marketplace involved. Here is how it works, when to use it, and where the trust traps hide when the ZIP comes from someone else. https://studiomeyer.academy/playbooks/claude-code-plugins-aus-zip-und-url-laden 30 Claude Projects for non-coders, a productive workspace in 30 minutes Projects in claude.ai is not a code feature. It is the workspace where you plan recurring tasks, store knowledge and train Claude without writing a line of code. 10 steps from empty project to productive setup. https://studiomeyer.academy/playbooks/claude-projects-fuer-nicht-coder 31 Build a prompt library, reuse your favourite prompts without copy-paste You build a stock of 20 to 30 prompts that carries your daily routine. 10 steps from the first entry to a searchable library that works with Claude.ai, ChatGPT and Claude Code. https://studiomeyer.academy/playbooks/prompt-bibliothek-aufbauen 32 Output Styles for Claude Code, your own tone in 20 minutes How to give Claude Code the answer personality you need via custom Output Styles. One Markdown file per style, frontmatter with name and description, project-level or global. Includes the traps I collected on my first attempt. https://studiomeyer.academy/playbooks/output-styles-fuer-claude-code 33 Statusline for Claude Code, context at a glance in 15 minutes How to build your own statusline with one shell script line in settings.json. Current model, context percentage, Git branch, cost ticker. With a concrete example script and the traps I collected on my first attempt. https://studiomeyer.academy/playbooks/statusline-fuer-claude-code 34 Claude Code checkpoints and /rewind: your safety net for risky edits How to undo broken AI edits in seconds with /rewind and checkpoints, free up context and try side quests without fearing for your code https://studiomeyer.academy/playbooks/claude-code-checkpoints-und-rewind 34 Keeping discovery files consistent so AI crawlers do not quote old nonsense bios agents.json, llms.txt, MCP server bios and agent knowledge files drift away from your website. When an AI crawler pulls the wrong file, pure marketing fiction about you ends up in Grok or ChatGPT. Ten steps to an audit. https://studiomeyer.academy/playbooks/discovery-files-konsistent-halten 34 Debugging hooks when nothing fires Ten steps to work out why your Claude Code hook never kicks in. From the JSON typo to the matcher pattern. https://studiomeyer.academy/playbooks/hooks-debuggen-wenn-nichts-feuert 35 Memory backup before an update, so you do not cry over a schema migration bug How to secure your agent memory, your Claude Code configs and your MCP server state before you update or migrate. With a concrete pg\_dump command, a restore test and a retention strategy. https://studiomeyer.academy/playbooks/memory-backup-vor-update 36 AGENTS.md alongside CLAUDE.md, one repo context for several AI tools Why you should maintain AGENTS.md in addition to CLAUDE.md when your team works with Claude Code, Cursor, Codex or Copilot in parallel. Over 60,000 repos already use the standard. https://studiomeyer.academy/playbooks/agents-md-neben-claude-md 47 Generative Engine Optimization for small sites, so Perplexity and ChatGPT-Search cite you Solo site, service site or portfolio. You want AI assistants to find and cite you when someone asks about your topic. Ten steps from inventory to citation tracking. https://studiomeyer.academy/playbooks/geo-fuer-kleine-sites 51 Background sessions with Claude agents: run long tasks without blocking your terminal In 20 minutes you dispatch Claude Code as a background session, let long tasks run in the background and pull them back later with resume. Including worktree isolation, model and effort defaults, and a SessionStart hook for context. https://studiomeyer.academy/playbooks/background-sessions-mit-claude-agents 52 Set up the Claude Code bash sandbox: isolate commands instead of blindly allowing them In 20 minutes you set up the bash sandbox so Claude Code runs shell commands in an isolated environment. No more dangerously-skip-permissions, and the work still runs through. https://studiomeyer.academy/playbooks/claude-code-bash-sandbox-einrichten 55 The expensive-model myth: when the top AI model is really worth it An honest decision guide for when you need the top class and when the cheaper one is enough. With concrete examples from coding, writing, agents and routine tasks. https://studiomeyer.academy/playbooks/mythos-klasse-wann-lohnt-sich-das 56 Setting up fallback models so Claude Code doesn't stall on overload When Opus is overloaded, Claude Code should switch to Sonnet automatically instead of aborting. With the fallbackModel setting you build a chain of up to three models. Ten steps. https://studiomeyer.academy/playbooks/claude-code-fallback-modelle 58 Security-guidance plugin for Claude Code, automatic security review while you build How to install and enable the official security-guidance plugin and feed it your own project rules, so Claude checks its own code changes for vulnerabilities and fixes them in the same session. https://studiomeyer.academy/playbooks/claude-code-security-guidance-plugin 59 Claude Code acting up after a config change, how to find the cause With the new safe mode you start Claude Code without all your customizations and switch them back on one by one until you find the culprit. Ten steps against the maddening error. https://studiomeyer.academy/playbooks/claude-code-safe-mode-diagnose 60 MCP is stateless: what the 2026-07-28 spec changes and what you need to do now The biggest MCP revision since launch went final on 28 July 2026, and the new TypeScript packages are at v2.0.0. No handshake, no session, new headers. What it means for your servers. https://studiomeyer.academy/playbooks/mcp-spec-2026-07-was-aendert-sich 66 Tracking model availability, when a model vanishes overnight Models come and go faster than you think. How to build a fallback chain, actively check availability, and switch over without breaking production. Ten steps. https://studiomeyer.academy/playbooks/modell-verfuegbarkeit-tracken 68 Using Auto Mode in Claude Code safely, in 10 steps without suicide commands Auto Mode lets Claude Code approve its own tool calls, with safety checks in the background. What it blocks, how to turn it on, which deny rules and worktrees give you a second net, and when to keep your hands off. https://studiomeyer.academy/playbooks/claude-code-auto-mode-sicher-nutzen 72 Effort levels in Claude Code, when low is enough and when you need max Effort controls how deeply Claude thinks, independent of the model. The five levels plus ultracode, how to switch live, set it persistently and pick the right one per task. https://studiomeyer.academy/playbooks/effort-level-in-claude-code-einstellen 74 Fact-checking an AI answer: 10 steps against wrong information How to check in a few minutes whether the AI is giving you facts or inventing them. A hands-on routine, no technical background needed. https://studiomeyer.academy/playbooks/ki-antwort-faktenchecken 76 AI for freelancers, writing quotes and client emails in half the time Quotes, follow-up emails and rejections eat hours every week when you're solo. With a fixed AI setup you write them in 10 to 15 minutes instead of an hour, without sounding like a template. 10 steps from empty chat to a reliable writing assistant. https://studiomeyer.academy/playbooks/ki-fuer-selbststaendige-angebote-und-mails 78 AI meeting notes that are actually usable How to let AI transcribe, summarize and turn meetings into tasks without ending up with an unreadable transcript. For solo founders and anyone stuck in too many calls. https://studiomeyer.academy/playbooks/ki-notizen-fuer-meetings 79 Local AI on your own machine, so your data never leaves the house You want to use AI but not send every input to someone else's server. Local models run entirely on your machine, offline, no account. Ten steps from the hardware question to your first chat, with and without a terminal. https://studiomeyer.academy/playbooks/lokale-ki-auf-dem-eigenen-rechner 82 What AI tools really cost, and where the hidden bills hide Subscription, API, free tier, token costs. Understanding the cost model behind AI before a bill surprises you at month's end. For everyone who works with AI instead of just playing with it. https://studiomeyer.academy/playbooks/was-kostet-ki-wirklich 95 Framing a deep research brief so the report is actually worth something Deep research hands you a report with dozens of sources. Whether it is usable is decided in the two minutes beforehand. Ten steps from a vague keyword to a brief that produces a result you can genuinely use. https://studiomeyer.academy/playbooks/deep-research-auftrag-formulieren 96 Analysing numbers and tables with AI without ending up with nonsense Revenue list, customer export, timesheets. AI builds you an analysis in minutes, but it computes quite differently from a spreadsheet. Ten steps for numbers you can trust instead of plausible-sounding inventions. https://studiomeyer.academy/playbooks/zahlen-und-tabellen-mit-ki-auswerten 98 AI as a sparring partner, better decisions instead of polite agreement AI likes telling you what you want to hear. That is exactly what makes it a poor adviser and a good sparring partner, if you set it up right. In 10 steps you get Claude or ChatGPT to pull your decision apart instead of nodding it through, expose assumptions, build the counter-position and find blind spots. https://studiomeyer.academy/playbooks/ki-als-sparringspartner-fuer-entscheidungen Build it yourself 03 Your first MCP server in 90 minutes No theory, no protocol-spec chapter. You build a server, you connect it, you watch it run. https://studiomeyer.academy/playbooks/erster-mcp-server-in-90-minuten 06 Your first sub-agent in 30 minutes How to write a small specialist sub-agent in Claude Code that takes one task off your plate, instead of prompting the same thing manually every time. https://studiomeyer.academy/playbooks/erster-sub-agent-in-30-min 07 Publishing an MCP server step by step You've built an MCP server. Now someone else has to be able to install it without asking you. Here are the ten steps from 'works on my laptop' to 'npx -y your-server'. https://studiomeyer.academy/playbooks/mcp-server-publishen 10 Scheduled agents with Claude Code routines Routines are cloud-hosted scheduled agents. Here is how to build a daily brief, an inbox triage, or a deploy watcher as a routine without your own server. https://studiomeyer.academy/playbooks/scheduled-agents-mit-routines 11 Avoiding tool sprawl, from 530 to 30 MCP tools per session Once you have more than 5 MCP servers connected, your agent starts to drift. 10 steps to clean tool routing per task. With the pickMcp pattern, hosts filtering and progressive disclosure. https://studiomeyer.academy/playbooks/tool-sprawl-vermeiden 12 Structuring your subagent team, the six roles you need in a daily driver Once your first subagent works, the next question hits: who does what on the team? A concrete role split for solo founders and small teams, with permission profiles and conflict avoidance. https://studiomeyer.academy/playbooks/subagent-team-strukturieren 14 Your own slash commands for Claude Code, productive in 30 minutes How to equip Claude Code with your own /review, /test, /deploy commands. One markdown file per command, done. With frontmatter, arguments, file references and bash execution. https://studiomeyer.academy/playbooks/eigene-slash-commands-fuer-claude-code 15 Claude Code headless in CI/CD, a setup that runs tomorrow claude -p is the workhorse mode for scripting. We build a GitHub Actions workflow that checks tests on every PR and auto-fixes lint errors, without anyone watching. https://studiomeyer.academy/playbooks/claude-code-headless-in-ci-cd 16 Agent eval in 60 minutes, how to find out if your agent is any good You built an agent. But does it work? In 60 minutes you build a small eval setup that tells you at the push of a button whether new changes make the agent better or worse. https://studiomeyer.academy/playbooks/agent-eval-in-60-minuten 21 Opus 4.7 migration, what changed for API builders Opus 4.7 has been GA since April 16, 2026 and brings three breaking API changes plus a new xhigh effort level as the Claude Code default. If you have code running against 4.6, here is what you need to update. https://studiomeyer.academy/playbooks/opus-4-7-migration 22 Your first Skill in 30 minutes, when slash commands aren't enough Skills trigger Claude automatically on natural-language phrases instead of /commands. You write a Markdown file, Claude decides when to load it. With frontmatter triggers, allowed-tools and bundled scripts. https://studiomeyer.academy/playbooks/erste-eigene-skill-in-30-min 23 Managing the context window in long Claude Code sessions When to /compact, when to /clear, when to /resume. Ten steps against drift in hours-long coding sessions. https://studiomeyer.academy/playbooks/context-window-managen-claude-code 24 Advisor Tool, Opus as advisor and Sonnet as executor in one API call The Advisor Strategy in the Claude API. A fast executor model drives the task, a stronger advisor model is consulted only when the executor gets stuck. Beta header, tool schema, valid model pairs, cost model, three use cases. https://studiomeyer.academy/playbooks/advisor-tool-opus-als-berater-sonnet-als-executor 26 Distributing plugin bundles via URL, --plugin-url in 30 minutes Since Claude Code 2.1.129 (April 2026) you can fetch plugin ZIPs directly from a URL. No marketplace entry, no repo setup. How the flag works, when it beats the marketplace path, and where the trust pitfalls hide. https://studiomeyer.academy/playbooks/plugin-bundle-via-url-distribution 28 Your first Claude Code plugin in 60 minutes, from empty folder to installable bundle Plugins bundle slash commands, skills, hooks, sub-agents and MCP servers into one unit you can share. Here are 10 steps from empty folder to a plugin your colleague pulls in via /plugin install. https://studiomeyer.academy/playbooks/dein-erstes-claude-code-plugin 34 Parallel Claude Code sessions with git worktrees, two branches at once without conflict How to run several Claude Code sessions in parallel with --worktree, without your feature and your bugfix overwriting each other's files. 10 steps with config, .worktreeinclude and subagent isolation. https://studiomeyer.academy/playbooks/parallele-claude-sessions-mit-worktrees 34 Skill does not trigger, what to check in 10 steps before you type /command again Skills are supposed to load automatically when Claude recognises they fit. Sometimes they do not. Here is the path I walk, from frontmatter check to settings.json, with concrete examples from real skills I have built. https://studiomeyer.academy/playbooks/skill-trigger-debugging 34 Sub-agent or skill or MCP tool, which one do you pick when Three ways to extend Claude Code that look almost identical but work completely differently. A decision guide so you don't build the wrong thing and end up migrating three days later. https://studiomeyer.academy/playbooks/sub-agent-oder-skill-oder-mcp-tool 35 Running parallel Claude Code sessions with git worktrees How to work on several branches at once with Claude Code without ending up in stash hell. One worktree per branch, one session per worktree, no switching. https://studiomeyer.academy/playbooks/claude-code-mit-git-worktrees 50 Opus 4.8 migration, the first Opus update with no breaking changes Opus 4.8 has been GA since May 28, 2026. Unlike the 4.7 jump there are zero breaking API changes, you just swap the model field. What still changes in behaviour, plus two new levers: Fast Mode and Dynamic Workflows. https://studiomeyer.academy/playbooks/opus-4-8-migration 53 From script to your own agent: the Claude Agent SDK in 90 minutes You know Claude Code as a CLI. With the Agent SDK you build the same agentic loop into your own code, with tool control, budget limits and MCP integration. Step by step. https://studiomeyer.academy/playbooks/claude-agent-sdk-erster-eigener-agent 54 Build an MCP App: interactive UI in your MCP tool instead of a wall of text How to give your MCP server a dashboard, a form or a map that renders right inside the chat. With the official ext-apps package, step by step. https://studiomeyer.academy/playbooks/mcp-app-interaktive-ui-bauen 57 Dynamic workflows in Claude Code, your first orchestrated multi-agent script What dynamic workflows are, when to use them instead of subagents, and how to start, watch, save and keep your first workflow under control in a single session. https://studiomeyer.academy/playbooks/dynamic-workflows-in-claude-code 61 Your first real n8n flow: automation for people who don't code From idea to running workflow. One trigger, one action, one notification, without a single line of code, in under an hour. https://studiomeyer.academy/playbooks/mein-erster-n8n-flow-fuer-nicht-coder 62 Agentic payments with x402, what builders need to know about machine-paid APIs x402 revives the old HTTP status 402 and turns it into a payment layer for AI agents. In 10 steps you understand the flow, build a paid API and let an agent pay autonomously, without an account and without checkout. https://studiomeyer.academy/playbooks/agentic-payments-x402 63 GPT-5.6 or Opus 5, which model for which task The honest decision guide without vendor-fanboy waffle. Ten steps from tier thinking through cost and context to your own mini-eval. By the end you know when the expensive model is worth it and when it isn't. https://studiomeyer.academy/playbooks/gpt-5-oder-opus-4-8-welches-modell 64 Subagents that spawn subagents, Claude Code's 5-level rule Since June 2026 subagents can start their own subagents, up to five levels deep. When depth really helps, when it blows up your cost and context, and how to set a depth budget. https://studiomeyer.academy/playbooks/subagents-die-subagents-spawnen 65 Setting up Claude Code agent teams, multiple sessions working together Agent teams let several Claude Code instances work on a task in parallel, with a shared task list and direct communication between them. Different from subagents. Setup in 10 steps, with the exact behavior and the cases where a single agent is the better choice. https://studiomeyer.academy/playbooks/claude-code-agent-teams-einrichten 70 Build an AI portfolio that lands you a job, in 10 steps As a career changer you don't need three years of experience, you need three projects someone can look at. This guide shows step by step how to turn an idea into a visible portfolio. https://studiomeyer.academy/playbooks/ai-portfolio-fuer-den-job 71 Proving your AI skills credibly in CV and interview You can work with AI, but in hiring everyone sounds the same. This guide shows how to back up your skills so they read as substance, not buzzword bingo. https://studiomeyer.academy/playbooks/ai-skills-belegen-lebenslauf-interview 75 AI images for your brand, from first prompt to finished set How solo founders create on-brand images with AI without a designer. Tool choice, prompt scaffold, a consistent look, product mockups and what you have to label legally. https://studiomeyer.academy/playbooks/ki-bilder-fuer-deine-marke 80 Understanding the three MCP primitives. Tools, Resources and Prompts Almost every MCP server uses only tools. But there are three building blocks, and knowing all three lets you build better servers and instantly grasp what a foreign server actually offers. In 20 minutes you'll have the model in your head. https://studiomeyer.academy/playbooks/mcp-primitives-tools-resources-prompts 81 Reliable JSON from any AI, so your automation doesn't break Why AI answers crash your automation and how to force a JSON format in ten steps that parses every time. Provider-independent, with ready-made prompt building blocks. https://studiomeyer.academy/playbooks/verlaessliches-json-aus-ai 85 Opus 5 migration, when thinking becomes the default Opus 5 landed on July 24, 2026. The model name is one line, the rest is not: thinking now runs by default, max\_tokens counts it, and one old combination throws a 400. Plus the prompt sentence you should delete today. https://studiomeyer.academy/playbooks/opus-5-migration 87 Stopping agents that drift off course Four patterns for how an agent derails mid-run, and ten steps that catch it before it works two hours in the wrong direction. https://studiomeyer.academy/playbooks/agenten-die-abdriften-stoppen 88 Seeing what your agent actually did, tracing in 10 steps with OpenTelemetry An agent runs, makes twenty tool calls and delivers something at the end. Why? Tracing opens the black box and shows you every step on a timeline. Here is how to set it up. https://studiomeyer.academy/playbooks/agent-tracing-mit-opentelemetry 89 Having AI summarize long documents without it inventing things A 40-page contract, a letter from an authority, a report. You want the short version, but not the hallucinated one. How to get a summary you can rely on, in 10 steps. https://studiomeyer.academy/playbooks/lange-dokumente-sicher-zusammenfassen 94 Why your screenshot shrinks, and how to control it The mechanics behind image understanding: resolution tiers, visual tokens, the cropping trick, PDF limits and the five documented things image understanding cannot do. For people who already know the upload button. https://studiomeyer.academy/playbooks/bilder-und-screenshots-an-die-ki-geben 97 Reading AI benchmarks without being taken for a ride Every AI company claims its model beats all the others. Ten steps for putting the numbers in the press release into perspective, spotting the usual tricks and finally building your own mini test, the only one that counts for your work. https://studiomeyer.academy/playbooks/ki-benchmarks-richtig-lesen 99 Learning a new skill with AI, the honest way AI is the most patient tutor in the world, and that is exactly the danger. Ten steps for actually learning something instead of just believing you can do it. For career changers, lateral entrants and anyone teaching themselves something alongside a job. https://studiomeyer.academy/playbooks/mit-ki-neue-faehigkeit-lernen 100 Changing your MCP server without breaking your users Your server is published and you want to rework a tool. In 10 steps you make the surface visible, check changes automatically and roll over without other people's setups quietly breaking. https://studiomeyer.academy/playbooks/mcp-server-aendern-ohne-nutzer-zu-brechen Local AI systems 101 Your first local AI model in 30 minutes A language model on your own machine, no cloud and no account. Ten steps from install to a first useful result, including the places where most people give up and the settings that still call out even with a local model. https://studiomeyer.academy/playbooks/erstes-lokales-modell-in-30-minuten 102 Which local model fits your hardware From eight gigabytes to a server with two graphics cards. How to get from your actual kit to the right model size in ten steps, instead of guessing and then learning from the crash. https://studiomeyer.academy/playbooks/welches-lokale-modell-passt-zu-deiner-hardware 103 Local AI and the GDPR, what it really solves and what it does not Local does not automatically mean compliant. Ten steps through the question of which privacy problems your own machine actually removes, which remain, and what a setup looks like that survives an audit. https://studiomeyer.academy/playbooks/lokale-ki-und-die-dsgvo 104 MCP with local models, tools without the cloud A local model that does not just talk but reads files, checks the calendar and uses your memory. Ten steps to a fully self-owned setup, including the point where small models fail at tool calling. https://studiomeyer.academy/playbooks/mcp-mit-lokalen-modellen 105 Local or cloud, the honest arithmetic When your own hardware pays off and when it is an expensive hobby. Ten steps through a calculation that counts electricity, your working time and the quality gap, instead of just putting hardware price against token price. https://studiomeyer.academy/playbooks/lokal-oder-cloud-die-ehrliche-rechnung 107 Transcribing conversations offline, without them leaving the building Client meeting, medical history, job interview. Ten steps to a transcription that runs entirely on your own machine, including the two conditions without which the promise does not hold. https://studiomeyer.academy/playbooks/gespraeche-offline-transkribieren 108 Connecting n8n to a local model Automation where the data stays in the building. Ten steps from the networking question through the five Docker constellations to the node that quietly cannot do tools and therefore never works with the agent. https://studiomeyer.academy/playbooks/n8n-mit-lokalem-modell-verbinden 109 Reading documents offline when the scan must not be uploaded Medical letter, contract, personnel file. Ten steps to text recognition that stays on your machine, including the distinction between a tool that misreads characters and one that makes up whole facts. https://studiomeyer.academy/playbooks/dokumente-offline-auslesen Workflow 09 Git for AI, 30-minute quickstart Step-by-step setup when you build with Claude or Cursor. From GitHub account to working branch workflow plus the GitHub MCP server. Bonus: what to do when the AI breaks something again. https://studiomeyer.academy/playbooks/git-fuer-ki-quickstart 29 Pull request reviews with Claude Code, 10 steps for a PR workflow nobody blocks anymore Pull requests are where AI code either slips through silently or blows up the build. Here are 10 steps to use Claude Code as a second pair of eyes for reviews, without your reviewer colleague losing their work or paralysing the pipeline. https://studiomeyer.academy/playbooks/pull-request-reviews-mit-claude-code 73 From Academy skill to your first paid AI gig You've finished the levels, you can handle MCP, memory and Claude Code. Now what? This playbook takes you in ten steps from learned skills to the first person who pays you for them. Honest, no cold-calling fairy tales, with concrete numbers. https://studiomeyer.academy/playbooks/erster-bezahlter-ki-auftrag 93 Using AI without losing your own edge Working with AI every day makes you faster. Research also shows your own judgement suffers, and you don't notice it. Ten steps for keeping both. https://studiomeyer.academy/playbooks/ki-nutzen-ohne-selbst-zu-verlernen Security 20 MCP STDIO security, what the OX Security story means for you In April 2026 OX Security disclosed a systemic RCE vulnerability in the MCP STDIO transport. 11 CVEs, up to 200K vulnerable instances. Here is what you actually need to do as an operator and as a builder. https://studiomeyer.academy/playbooks/mcp-stdio-sicherheit 25 Secure your MCP server with OAuth 2.1 in 60 minutes When you need auth for your MCP server and when you don't. PKCE mandatory, .well-known discovery, Resource Indicators (RFC 8707). Three setup options: WorkOS AuthKit, Auth0 for MCP, or your own authorization server. https://studiomeyer.academy/playbooks/mcp-server-auth-mit-oauth-2-1 28 Confused deputy audit for Claude Code, done in 25 minutes May 2026, four security teams found the same failure pattern in Claude Code, Chrome extensions and apps within 48 hours. Here is the audit matrix broken down into ten concrete checks for your own setup. https://studiomeyer.academy/playbooks/confused-deputy-audit-fuer-claude-code 83 Understanding prompt injection, the three traps when your AI assistant reads foreign text Once your assistant reads emails, PDFs or web pages, foreign text can slip it commands. Not a coder problem. Here are the three places it happens and the three habits that protect you, worked through in twenty minutes. https://studiomeyer.academy/playbooks/prompt-injection-erkennen 84 Protecting yourself from AI fraud, fake calls, phishing and bogus job offers Voice cloning, deceptively real phishing emails and fake recruiters. How to spot the new AI scams and protect yourself in ten concrete steps. No tech degree required. https://studiomeyer.academy/playbooks/dich-vor-ki-betrug-schuetzen 90 Vetting third-party MCP servers safely before you install them An MCP server gets more power over your data than any app you have ever installed. These 10 steps are the check you run in five minutes before clicking 'Add'. https://studiomeyer.academy/playbooks/fremde-mcp-server-sicher-pruefen 106 When a stranger's repo configures your AI. The ChainDrop case In August 2026 a worm started writing startup hooks into AI agent config files. The trigger isn't installing, it's starting your tool inside the project. Ten steps to check, respond and harden for good, including the question of what order to clean up in. https://studiomeyer.academy/playbooks/agent-konfiguration-als-angriffsweg Law & privacy 05 DACH legal. EU AI Act and GDPR for AI operators What's been binding since February 2025, what gets enforced from August 2026. What SMBs and freelancers must implement now. No legalese, just duties. https://studiomeyer.academy/playbooks/dach-legal-eu-ai-act-und-dsgvo 67 Data protection when working with AI tools: ten habits that keep you safe What never belongs in a chat window, how to set the training opt-out, and when local models are the better call. Practical, no legalese. https://studiomeyer.academy/playbooks/datenschutz-mit-ki-tools 77 Spotting and correctly labelling AI content How to recognize AI-generated images, text and voices and when you have to label your own AI content from August 2026. Explained practically for solo founders with no legal department. https://studiomeyer.academy/playbooks/ki-inhalte-erkennen-und-kennzeichnen 86 The AI policy for small teams, on a single page Ten steps to an internal rule for two to ten people that actually gets read and followed. With a template you can copy. https://studiomeyer.academy/playbooks/ki-richtlinie-fuer-kleine-teams Was als nächstes ### Playbooks bringen Dich in Gang. Die Kurse machen es tief. Playbooks sind Einzel-Anleitungen. Wenn Du systematisch von Grundlagen bis eigene MCP-Server lernen willst, gibts die Academy in sechs Leveln. Level 1 bis 3 sind kostenlos und zeigen die Grundlagen. Ab Level 4 wirds Memory-First und Multi-Agent. [Zu Level 1](https://studiomeyer.academy/levels/1) [Assessment machen](https://studiomeyer.academy/start) ## Sitemap Public pages of this site as Markdown: [sitemap](https://studiomeyer.academy/sitemap.md). Curated entry point: [llms.txt](https://studiomeyer.academy/llms.txt), full text: [llms-full.txt](https://studiomeyer.academy/llms-full.txt).