Cloudron Forum

Imho it is best to copy the librechat.example.yaml from here https://github.com/danny-avila/LibreChat/blob/main/librechat.example.yaml into your librechat.yaml and then copy the examples for your custom AI endpoints into it (eg https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints/mistral). Btw: openai & gemini also work for me without librechat.yaml. [image: 1751800848541-86ea849f-7d98-4a60-b726-3beb00af3e6d-image.png]

Hello @shrey This is expected. Each Cloudron release and update has to treat every system the same and can not respect changes made by the user. Changes made to the core services need to be managed by the user manually.

@girish Thanks for your explanation, I will add version within the The appStoreId, @james so now I have something pretty robust and will use the following process: All updates disabled by default I will manually check and update Cloudron quarterly, on specific timeframes I will leverage on API to orchestrate each app upgrade individually

Connecting... root@16405153-e269-41e3-ab8d-095606d5b07e:/app/data/public/KNEL/FetchApply# gup Already up to date. root@16405153-e269-41e3-ab8d-095606d5b07e:/app/data/public/KNEL/FetchApply# cat /app/data/.bashrc cd /app/data/public/KNEL/FetchApply alias gup='su www-data -c "git pull"' root@16405153-e269-41e3-ab8d-095606d5b07e:/app/data/public/KNEL/FetchApply# Ah this opens up all manner of creature comforts. Excellent! Really appreciate the prompt response @joseph

I will put in a feature request now and we can discuss further in that thread.

Thank you @James In the interim, I am attempting a workaround <can't find fingers-crossed emoji>

OK - Internally, this went from "looks cool" couple of days ago to "must have" a few moments ago. Beside with my unlimited gratitude and limited packaging skills and experience, is there any other way I can help make Twenty on Cloudron happen sooner than later?

Great question! Outline does not support end-to-end encryption (E2EE) or client-side encryption, even in the self-hosted version. The encryption it uses — TLS for data in transit and AES-256 for data at rest — applies at the server level, meaning the server (whether cloud-hosted or self-hosted) still has access to the unencrypted document contents. So, while data is securely transmitted and stored, it is not encrypted in a way that prevents server access, which is what true E2EE or client-side encryption would provide.

@andirahmat oh, it fixed itself?

Thanks @joseph for referenced those links that I missed. So, if I'm not wrong, import will be made directly from Notion servers, not from Notion backup file, right?

Bundling the RAG API is in the works

@jdaviescoates said in Best privacy chat apps: VC mentality? Signal is a non-profit. No hard feelings. When something is backed by the wealth of a top 100 US fortune then saying that it's a non-profit is like playing the violin to me. I don't doubt that it is a non-profit but it just sounds like branding or and opportunistic decision. Maybe I'm just sceptical. VC and philanthropy go hand in hand but it's a societal model that I don't endorse. One can of course focus on the tech and its promises. Then Signal clearly stands out compared to other products. But like the point about encryption, I disassociate "best" and "privacy" a little (topic title). I absolutely belive that no useful Signal data can be retrieved from the AWS and Azure server, but nevertheless every user indirectly funnels money to these two unethical (at minima) corporations. Once it's sufficiently private (a fundamental right), let's look at the broader picture as well. I see tech is a cultural product and the adjective "best" can be used in a holistic sense. Just wanted to give a heads up for Threema in that regard. Move slow and fix things ️

@nebulon said in Open AI Integration: you have to prefix the env variables with export statement, as env.sh is a .sh file. So like: export OPENAI_API_KEY=... Ah, I didn’t know that – thanks a lot for the reply! Problem solved