AI Library
The Double Structure of Digital Sovereignty
Europe’s Departure from Palantir and the Chains of American Big Tech
Kim Kyung-jin, Attorney at Law
This is a record of 2026, when European intelligence agencies and defense ministries began removing analytics tools from America’s Palantir. It covers the replacement decisions made by France’s General Directorate for Internal Security (DGSI), Germany’s Federal Office for the Protection of the Constitution (BfV), and the Netherlands Ministry of Defense; the incident in which US export controls severed an ally’s ac…
New English Edition
Artificial Intelligence in Horticulture
Kim Kyung-jin, Attorney at Law
Across five chapters and ten sections, this book examines computer vision for crop diagnosis, harvesting robots and autonomous field systems, smart greenhouses and digital twins, precision irrigation and supply-chain quality control, high-throughput phenotyping, and predictive breeding.
New English Edition
Artificial Intelligence in Food Crop Agriculture
Kim Kyung-jin, Attorney at Law
Across six chapters and eighteen sections, the book examines digital agricultural infrastructure, remote sensing, crop diagnosis, yield forecasting, precision irrigation, genomics, molecular breeding, agricultural robotics, climate-smart agriculture, and global food security.
New English Edition
The Future of Forestry and Agroforestry
Kim Kyung-jin, Attorney at Law
Driven by Artificial Intelligence and Digital Innovation
Across five chapters and fifteen sections, the book follows satellites, drones, LiDAR, digital twins, forest-specific language models, wildfire and pest forecasting, forestry robotics, agroforestry, timber traceability, and forest carbon markets.
New English Edition
Smart Livestock Farming: AI Enters the Barn
Kim Kyung-jin, Attorney at Law
Sensors listen, cameras watch, and artificial intelligence helps farmers decide.
Across five chapters and fifteen sections, the book follows precision livestock farming from animal health and reproduction to robotic milking, virtual fencing, digital twins, methane reduction, welfare, and data ownership.
Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] Chapter 27: Beginner Hacks: 10 Essential Tips
Mastering Claude Code
Chapter 27: Beginner Hacks: 10 Essential Tips
Kim Kyung-jin
Mastering Claude Code
The First Encounter with Speed
Recall the day you first opened Claude Code in the terminal. Standing before the black screen with a blinking cursor, you hesitated for a long moment, unsure what to type. You entered a prompt, received output, fixed it, then fixed it again. Hours passed and you still did not have what you wanted. Yet someone else using the same tool produced results with striking speed.
It is not the tool that differs. Only a handful of habits.
The ten tips introduced in this chapter are fundamentals that anyone new to Claude Code must learn first. Each is a small habit, yet when they accumulate, your work speed and output quality shift noticeably.
Tip 1. Initialize Your Project with /init
Suppose you open a project folder with existing files in Claude Code. The first thing to do is type the /init command.
Running this command makes Claude Code scan your entire project. It identifies the folder structure, file list, and code conventions, then automatically generates a claude.md file. This file acts like a cheat sheet for your project. It contains information about how the architecture is organized, which files are core, and what rules apply.
Once you run /init, you do not need to re-explain the project in every session. Claude already has the context.
What if you are starting a new project from scratch? In that case, you build the claude.md file yourself while explaining the project goals, tech stack, and key rules. The better this file is prepared, the easier all subsequent work becomes.
Tip 2. Set Up Your Status Line
When you are deep in coding, it is easy to forget how much context you are consuming. You have probably had the experience of suddenly checking and finding that your context window is nearly full.
Type /statusline in the terminal and tell it what information you want to see. Claude Code creates a small dashboard at the bottom of your terminal. You can display the current model name, context usage percentage, cost, and more in real time.
The real value of the status line is that it lets you detect context depletion before it happens. If you keep working without knowing you are running low on context, Claude's response quality drops sharply. With a status line, you can prepare before that situation arrives. A small script like this governs the quality of your entire session.
Tip 3. Use Voice Input with /voice
Typing long prompts on a keyboard takes more time than you might think. You have specific requirements in your head, but the process of converting them to text often loses the core.
Claude Code has a native voice input command called /voice. You can give instructions directly to the terminal by speaking. Say 'improve the error handling in this function' as it comes to mind, and it becomes a prompt.
The advantage of voice input is not just speed. When you explain by speaking, context and intent naturally come through. Descriptions that you unconsciously trim when typing become richer in speech. As a result, Claude grasps your intent more accurately.
Using a separate voice dictation app is also a good approach. Through your operating system or third-party apps, you can input text by voice anywhere, raising efficiency not just in Claude Code but in other work as well.
Tip 4. Keep Your Context Small
The thought 'the more I put in, the more it will understand' is natural, but in Claude Code it becomes a poison.
Do not dump your entire codebase into the conversation. Provide only the information needed for the current task. Break large problems into small steps, and pass only the relevant code and context for each step.
The less noise in the context window, the higher Claude's performance. The same is true for humans. If you discuss ten items at once in a meeting room, none of them reach a proper conclusion. Deep results come only when you focus on one thing at a time.
Tip 5. Diagnose Token Consumption with /context
You were told to keep context small, but how can you know what is consuming tokens right now?
Use the /context command. This command shows you as a percentage where and how many tokens are being consumed in your current session. The proportion taken by the system prompt, the proportion taken by file content, and the proportion taken by MCP server definitions all become clear at a glance.
When your session feels heavy, when responses slow down or quality drops, run /context once. Often you will find tokens being consumed in large amounts in unexpected places. Once you identify the cause, you can restructure to free up capacity.
This diagnostic habit pairs with Tip 4, 'Keep Your Context Small'. Understanding a principle is one thing, but without measuring your current state, practice becomes difficult.
Tip 6. The 60% Compression Rule and Resetting Between Tasks
When context usage reaches around 60%, it is time to run the /compact command.
This command compresses your conversation history. It filters out unnecessary parts while preserving important information. You can continue working after compression without restarting the conversation from the beginning.
Here is a useful technique. When running /compact, you can explicitly specify what to preserve. For example, if you type '/compact, but keep the API integration decisions and database schema,' Claude will trim the rest while leaving the specified information intact.
When switching to a completely different task, use /clear. It wipes the conversation history clean and starts like a new session. Your claude.md file and project files remain, so it differs from starting from scratch. Making a habit of resetting with /clear each time your task changes prevents lingering context from the previous work from polluting the new task.
Tip 7. Prioritize Plan Mode
Press Shift+Tab to switch modes in Claude Code. Plan Mode is one where Claude can read and investigate code but makes no changes.
There is a reason this matters.
In Plan Mode, Claude lays out a work plan. It organizes which files need changes, the order of approach, and what additional points need checking. When needed, it asks questions about unclear areas.
After going through this process, when you exit Plan Mode and give the execution order, the number of times Claude modifies code in the wrong direction and has to reverse course drops noticeably. The difference between jumping straight into coding without a plan and designing first, then coding. The principle is the same for humans and AI.
Tip 8. Treat the Agent Like a Junior Developer
There is a world of difference between directly commanding 'implement this feature' and posing a problem: 'what approach would help reduce user drop-off?'
If you give Claude only specific commands, Claude does exactly as told. But when you pose a problem, Claude reasons for itself, compares options, and explains why it chose a particular approach.
Two benefits arise from this process. First, because Claude builds its logic before writing code, the quality of the output improves. Second, as you review Claude's reasoning, you may discover considerations you had not thought of.
Think back to when you entrust work to a junior developer. If you only say 'Do it this way,' that person does not grow. If you ask 'How might we solve this problem?' that person thinks, and you can review that thinking. Claude is no different. When you give Claude a chance to think, it produces better results.
Tip 9. Question-Driven Prompting
Claude creates questions on its own in plan mode, but there are ways to use this more actively.
You add this instruction to the prompt: 'Keep asking questions until you are 95% confident about what I want and what you need to do.' With this, Claude works through ambiguous areas one by one. Who is the target user? What behavior do we expect when an error occurs? What is the performance standard? Requirements we failed to spell out become visible through questions.
What happens without this alignment process? Claude makes assumptions on its own and proceeds with the work. When you receive the result, you find three or four places where it differs from your intent. You request a revision, and revise again. If this back-and-forth repeats just three times, context burns through quickly.
Question-driven prompting blocks this back-and-forth cost in advance. If you invest five more minutes upfront to clarify requirements, you can save thirty minutes later.
Tip 10. Inserting Self-Validation into Your To-Do List
Claude Code creates a to-do list when it starts a task. The tenth tip is to embed validation steps directly into this list.
For example, if the task is building a website, you can structure the to-do list like this.
1. Implement website layout 2. Take a screenshot of the website and confirm the layout came out as intended 3. Open Chrome DevTools and check for functional errors 4. Confirm responsive design works correctly on mobile 5. Revise based on findings
When you do this, Claude does not just build something and hand it over to us. Instead, Claude reviews what it made, fixes problems if it finds any, and then shows us the result.
There is one additional technique. You tell Claude: 'Before moving to the next to-do item, repeat until you are 95% confident the current item is finished.' AI rarely produces perfect results on the first try, but running through a self-validation loop lifts a 60% result up to 90%. That difference compounds, and the overall project quality shifts noticeably.
Small Habits, Big Differences
We have looked at ten tips one by one. Start with /init, track progress with the status bar, give instructions quickly via voice, keep context light, diagnose and compress, plan first, raise questions, prompt for deeper thinking, and automate validation. None of these are complex techniques. They are simply habits.
Once this foundation becomes second nature, you are ready to move to the next level. More sophisticated methods await,parallel work using secondary agents, custom skill file creation, cost optimization.
Artificial Intelligence Expert Attorney Kyung-jin Kim
AI policy and law specialist. Former National Assemblyman. Multiple published works.
If this book stayed with you for even a moment, please support the next story that will come into the world.
(Voluntary donation account: NH Bank 302-1096-0948-81 Account holder: Kim Kyung-jin)
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.