as a planning tool. As a programming tool, it's really amazing and does an incredible job, and it's probably done me many times more than the most efficient developer.
However, Claude Code can also be used for many other very useful application areas. You can use it for tasks like making presentations, which are best done by writing code, for example Python or LaTeX. In addition, you can use Code Claude to use situations that are completely unrelated to coding, such as organizing your sales reach or building a knowledge base, which is what I will talk about in this article.
A powerful LLM knowledge base is an incredibly powerful concept that can make you more efficient simply because you can quickly retrieve relevant information. I'll cover how to set this up, why you should set it up, and how you can get the most out of your knowledge base.
Why establish a strong LLM knowledge base
First, I will cover why you should set up a strong LLM knowledge base. The simple reason is that LLMs work better when they have more context, and the more context you can give to your language models, the better able they will be to solve the different problems you throw at them.
Therefore, storing as much information as possible in a knowledge base is incredibly powerful because the LLM can access it when necessary.
I try to keep everything I do in a central database. This could be things like:
- the meetings I am a part of
- notes that I think about
- mistakes my agents make and how to avoid them
and so on.
Before LLMs, you would think there was no need to set up such a large knowledge base with so much information, because you couldn't just download the right information when you needed it. For example, if you need a note from a particular meeting, it can be a lot of work to find that particular meeting transcript first, and then, in that transcript, find the specific point you want and find the information you want.
However, this changed completely when large language models became a thing, because suddenly you can easily access a large amount of information in a few seconds. The availability of information has increased dramatically overnight.
The vast increase in LLM knowledge availability, making knowledge bases more important
To answer quickly, the reason you build a powerful LLM knowledge base is that it allows you to retrieve the right information when you or your coding agent needs it most. The fact that you gain a strong knowledge base through LLMs makes knowledge more accessible, which means you can retain a greater amount of knowledge.
How to set up an LLM enabled knowledge base
Now I will go into how to set up your LLM powered knowledge base. In fact, it's very simple. You just need to save all the information in one folder on your computer.
In fact, it's harder than you think because you need to make it a habit to keep all the information in that place. However, I will take you step by step on how I started setting this up, and by focusing on it over time I got better and better at updating my knowledge base and adding all the relevant information to it.
You can start by having someone take notes of the meeting and make sure that they always keep notes with all the important information, such as who attended the meeting, the time of the meeting, the name of the meeting, and the context of the meeting. This is information you can download from your calendar, for example
Just to add all your meeting notes and notes to, the central folder does most of the work.
Note also that the database does not have to be a local folder. It can also be cloud-based applications, like Notion, where you can save text files. The knowledge base doesn't really need to be more than that. It just needs to save the text.
In addition to this, I also have reminders several times a week to keep things in the knowledge base that I am thinking about, learning from my work, or other useful information that I have thought of. I try not to be or overthink what I add to the knowledge base. The most important thing is to add information, which I do by simply telling the Claude Code:
Add
on the basis of my knowledge
Claude Code then finds out which file or subfolder is most important to contain this information and saves the information for me.
Another thing I do to update my knowledge base is I ask Claude Code to do a daily walkthrough of all agent interactions throughout the day. This can be both interactions I've had with my personal Code Claude and I've had with my OpenClaw bots or other people I've had with my OpenClaw bots. We discuss what went well, what didn't go well, and it automatically saves the general information in my general information folder for later access. This is done by setting up a cron job, which runs daily.
Naturally you'll have a different workflow than I do, and thus keep different pieces of information and have access to different pieces of information as well. So I urge you to think about the different information you wish to keep, and try to organize that information into a knowledge base. A general rule of thumb is that you should keep as much context as possible in the knowledge base; It doesn't really matter if the context knowledge base is bloated, and you should try to make the storage of information in the knowledge base as automatic as possible.
What I mean by automatic is that you don't need, for example, to manually copy your meeting notes into a database every time you have a meeting. That is, a lot of manual work, which one can get annoyed to do later, and secondly, you may forget to do it after the meeting. Instead, you should set up an automated script or flow that stores the meeting notes in the knowledge base for you.
How to use a knowledge base
In the last section, I covered how to create a knowledge base and store information in it. The second part of the knowledge base is, of course, how you use the knowledge base in practice. With this, I have two high-level topics to cover.
- Search for pieces of information whenever you need it
- Give Code Claude or your other agents access to the information you can download if it's relevant to them to complete the job.
In many cases, I find myself looking for a piece of information that I know I discussed in a meeting or thought about in the past. In these cases, it is very worrying if you cannot access the information. I am asking Claude Code to look at my knowledge base and find an answer to my question. Sometimes it doesn't get a straight answer, but it does get some reasonable answers or pieces of information linked to my question, which is also very useful in many cases.
The second use of the knowledge base is to give Claude Code access to it, so that he can access it whenever he thinks it's appropriate. For example, if you ask them to complete a coding task, it may be useful information in their knowledge base. Whenever you make presentations, for example, you can look at previous presentations.
If the database is a central folder, you must ensure that, first, you have given Claude Code or your scripting agent access to that folder.
Second, you need to have a user-level skill file or claude.md file so that the coding agent knows about the knowledge base and how to access it at any time. It is important that your coding agent is aware of the knowledge base at any time, regardless of where you are removing the model.
Mistakes to avoid
I also want to cover some mistakes that are easy to make when creating a knowledge base. The first error I want to cover is what happens when the database becomes outdated. Of course, there is a drift in knowledge. Your opinions on topics may change, or parts of the information may be out of date. It is therefore important that you regularly check your information for outdated information.
This check could be a weekly cron job, where you have Claude Code go through all the interactions you've had in the past week and compare it to the knowledge base to look for outdated information.
Another common mistake is that you do not make the agent aware of the information base when it is working on certain folders. For example, if you only declare the knowledge base code agent within the project-level Claude.md file, Claude will not know the knowledge base if it is running in any other folder. This, of course, is very problematic, since when Claude works in that folder, he will not have access to all the same information.
This is where the user-level Claude files or skills.md files come in handy, as they are always loaded into your coding agent, no matter what folder you use them in.
The conclusion
In this article, I've covered how to build a Claude Code-powered knowledge base. This is a central storage for all the information you interact with every day, and it gives you a powerful advantage because you can access information quickly, and you can search the knowledge base using LLMs. Coding agents are incredibly efficient at searching large pieces of information, such as folders, and storing/accessing relevant information will be fast, even if the information base is large. I urge you to try to establish a personal information base. Try to keep all the information you interact with on a daily basis in that database, and try to use this knowledge base actively to search for information yourself and make it available to your coding agents to be more effective. I believe that such personal information based on preferences and specific information will become even more important in the future when coding agents become more powerful.
👋 Touch
👉 My Free eBook and Webinar:
🚀 10x Your Engineering with LLMs (Free 3-Day Email Course)
📚 Get my free ebook Vision Language Models
💻 My webinar on Vision Language Models
👉 Find me on social media:
💌 Stack
🐦 X / Twitter
