ChatCBT: AI-powered CBT journaling assistant for your notes

[clairefro]

ChatCBT is an AI-powered journaling assistant* for your Obsidian notes inspired by cognitive behavioral therapy (CBT)

Re-work negative beliefs by chatting with a kind and objective agent, powered by your choice of a cloud or local/private AI service. Generate a summary of your reframed thoughts to help change negative thinking patterns.

Conversations are saved as files on your local computer, which you can keep as a diary or share with a therapist.

Install here: Plugins - Obsidian

Grateful for your feedback

• Not a replacement for therapy. See disclaimer on “Learn more” page.

[Tom5]

That’s awesome. I tried installing it following the setup guide, however I found it keeps loading, and not finishing a command. Is there anything I’m doing wrong, or can do to find the error?

[Tom5]

I’m using Olama with a 8gb ram macbook, could that be the issue?

[Betsku]

Thank you for this! Works great, even in other language (finnish) I made bot to telegram to reframe my thoughts via custom cbt bot, but this helps me work inside obsidian and maybe journal more. Thanks!

[clairefro]

Hi @Tom5 , are you using the Ollama mode? if so, it could have to do with the specs of your laptop and how it handles a local LLM.

If you are using OpenAI - it may have to do with your internet connection as it is usually quite fast.

Edit: just saw your comment about using Ollama. Yes, that is probably why it is slow. I am working on a feature that let’s you pick which local model you’d like to use, which might help people with hardware limitations

If the issue is unrelated to either of the above, it may be a bug and I’d be curious to deep dive it

[clairefro]

In the newest version 1.0.3 you can now override the default models for OpenAI and Ollama with any model name of your choice

[s3880278]

Hi!
I have been trying to get this plug-in to work, but I am running into some issues. I have followed the instructions for how to use OpenAI as provider. I left model empty and am therfore using gpt-4o-mini. I have signed up fro OpenAI, created and filled in my API Key and added credit. However, when I run the chat command I get:

ChatCBT failed with status 401: Unable to connect to provider openai.

Ensure you have:

• a valid API key registered in the settings,
• there is credit charged into your account (if applicable)

I did try to use Ollama first, but the terminal work was a little too hard for my novice brain to figure out , so changed to OpenAI. Despite having provider set at OpenAI I am occasionally getting a pop-up saying

No Ollama model found. Install models via terminal: ollama pull model

Any idea what may cause this? Or more importantly, how I can fix it. Any terminal command I can use to “clean out” Ollama traces from Obsidian, and get OpenAi to work.

Thanks! And sorry if this is a really dumb question

[s3880278]

[clairefro]

hey there - are you still experiencing this issue? if so, can you share a screenshot of your settings page (you can blur the API key)