It's memory for Claude Code on your computer. I had to keep reminding Claude about the same things over and over. I wanted Claude to remember our previous conversation and the information that was part of those conversations. I wanted to be able to maintain a train of thought on a subject even after weeks. I also wanted Claude to understand how much time had passed between conversations. It was important to me that all the memory stayed securely local on my machine. It's mine and doesn't go anywhere. No Rekol accounts to create or API keys to store. It's also simple markdown, which can be searched so Claude can access it easily, time stamped and tagged to ensure memory is still valid and not stale. All you have to do is copy the install string and run it in a terminal, and then tell your Claude Code session to start using it. It helps Claude Code to work better so you still need whatever Claude Code needs to run. What I would like is to have as many people as possible try it for everyday use and provide feedback on ways I can improve it, just because it's been really useful for me, I wonder if it's actually useful for others and if there are ways to improve it or fix issues to make LLM use easier for everyday users.
I had an idea for a research project. But trusting an LLM that is constantly forgetting things is hard. It's like talking to a friend with amnesia. I've been in tech over 20 years, and AI is certainly changing the game. But having a research and coding partner that understands and remembers seemed important if I was to actually get it done. Context is limited but session history is not, so why do I have to keep reminding it things over and over. Everyone uses LLMs differently. I want to make Rekol, the memory that helps everyone, that's why your feedback is so important. How do you use LLMs and would memory make it easier?
It's memory for Claude Code on your computer. I had to keep reminding Claude about the same things over and over. I wanted Claude to remember our previous conversation and the information that was part of those conversations. I wanted to be able to maintain a train of thought on a subject even after weeks. I also wanted Claude to understand how much time had passed between conversations. It was important to me that all the memory stayed securely local on my machine. It's mine and doesn't go anywhere. No Rekol accounts to create or API keys to store. It's also simple markdown, which can be searched so Claude can access it easily, time stamped and tagged to ensure memory is still valid and not stale. All you have to do is copy the install string and run it in a terminal, and then tell your Claude Code session to start using it. It helps Claude Code to work better so you still need whatever Claude Code needs to run. What I would like is to have as many people as possible try it for everyday use and provide feedback on ways I can improve it, just because it's been really useful for me, I wonder if it's actually useful for others and if there are ways to improve it or fix issues to make LLM use easier for everyday users.
I had an idea for a research project. But trusting an LLM that is constantly forgetting things is hard. It's like talking to a friend with amnesia. I've been in tech over 20 years, and AI is certainly changing the game. But having a research and coding partner that understands and remembers seemed important if I was to actually get it done. Context is limited but session history is not, so why do I have to keep reminding it things over and over. Everyone uses LLMs differently. I want to make Rekol, the memory that helps everyone, that's why your feedback is so important. How do you use LLMs and would memory make it easier?