Learn AI Prompting vs Linux道場 − コマンドを目で見て覚える

Learn AI prompting the practical way. Short lessons show why one prompt gets a vague answer and a better one gets exactly what you need. It is a prompt engineering course on your phone: from your first clear request to system prompts, few-shot examples, chains and testing. No coding needed for the lessons. Every idea is explained in plain language, with before-and-after examples you can reuse with any AI chatbot or large language model (LLM). The same skills work across LLMs. WHAT'S INSIDE • 53 courses and 199 bite-sized lessons, from beginner (A1) to expert (C2) • 20 debugging missions: find out why a prompt failed and pick the fix • 20 guided projects in Python: templates, a JSON output check, a token budget and a small test set, step by step • AI mentor chat: ask about any concept and get a clear answer • Cheat sheet with 75 patterns, 50 quiz questions and 50 interview questions • Certificate exam when you finish the track • Streaks, XP and daily reminders • App in 22 languages WHAT YOU'LL LEARN Prompt writing basics • How to write prompts that name the task, the audience and the goal • Context, constraints, length and tone, so the model stops guessing • Output formats you can use: bullet points, tables and JSON • Iterative refinement: rewrite a weak request and compare the answers Prompt design patterns • Roles and system prompt design • Few-shot prompting: teaching the model with examples • Chain-of-thought and tree-of-thought reasoning • Reusable templates and variables • Extraction and classification • Temperature and sampling settings, and when they matter Advanced prompt engineering • Chaining and pipelines, self-improvement and LLM-as-judge • Testing, regression checks and debugging failures • Prompt injection and how to defend against it • Hallucinations, bias and guardrails for safer answers • Long documents, long context and retrieval-augmented generation (RAG) • Structured generation with JSON Schema • Code generation, image generation, vision, audio and voice • Tool use, AI agents and multi-agent orchestration • Automated optimization with DSPy, meta-prompting, and prompting vs fine-tuning DEBUGGING MISSIONS Each mission starts from a real problem: an answer that could be about anything, a model that still rambles after you said "be brief", a confident answer that was never in the document, three examples that produced three formats, or a request that breaks when a field is blank. You read the case, choose the fix and see why it works. Seven easy, eight medium and five hard. GUIDED PROJECTS Twenty short projects turn good habits into checks. You write small Python functions around a prompt: fill template slots, count tokens against a budget, parse a JSON reply, score answers against a rubric, redact personal data, and compare two versions honestly. Your code runs on a real Python runtime and prints the result. A simple stand-in replaces the model, so every run gives the same answer and the lesson stays clear. FREE AND PRO Free: the first course, 2 missions, the first project, one AI mentor question per day and one AI-generated lesson per week. The free version shows ads. Pro: every course and lesson, all 20 missions and 20 projects, unlimited AI mentor, the certificate exam, and no ads. Pro is a weekly, monthly or yearly auto-renewing subscription, or a one-time lifetime purchase. Payment is charged to your Apple Account when you confirm the purchase. A subscription renews automatically unless it is turned off at least 24 hours before the end of the current period, and your account is charged for renewal within 24 hours before the period ends. You can manage or cancel subscriptions in your App Store account settings. The skills work with any chatbot, including ChatGPT, Gemini and Claude. This app is not affiliated with OpenAI, Google or Anthropic. Terms of Use (EULA): https://www.apple.com/legal/internet-services/itunes/dev/stdeula/ Privacy Policy: https://www.coddykit.com/terms/privacy-policy

「lsとcdは分かる。でもパイプとリダイレクトで手が止まる」 そんな人のための、手を動かして学ぶLinuxコマンド入門アプリです。 ■ 打ったコマンドが、図になって流れる cat app.log | grep ERROR —— このとき何がどこへ流れているのか。 標準出力と標準エラーが別の線で描かれ、何行・何バイト流れたかまで見えます。 パイプとリダイレクトでつまずく理由は「見えないから」です。 ■ 39レッスンが完全無料 ・L1 移動とファイル操作(pwd cd ls mkdir cp mv rm) ・L2 検索とパイプ(grep wc sort head tail find | > >> 2>) ・L3 容量とアーカイブ(du df tar gzip・.tar.gz の往復) ・L4 パーミッション(chmod chown sudo ./script.sh) ・L5 プロセスと環境(ps kill jobs export PATH alias) 「command not found」や「Permission denied」が出る理由を、 失敗を先に体験しながら理解できます。 ■ 実務シナリオ編(追加コンテンツ) 「エラーログから原因を特定する」「ディスク逼迫を解消する」 「応答しないプロセスを止める」—— 新しいコマンドは出てきません。習ったものの組み合わせだけで、 現場のトラブルを調査から対処まで通しで解く8本。 さらに、毎回お題が変わる腕試しモード(ランダム出題)も解放されます。 ■ 4種類の練習環境(無料) 目標なしで自由に打てるサンドボックス。まっさら・小さなプロジェクト・ サーバーのログ・権限がこわれた の4種類から選べます。 ■ 完全オフライン 本物のLinux環境もSSHも不要。エラーメッセージまでbashとcoreutilsに合わせてあるので、 アプリで身につけた操作はそのまま実機で通用します。