Bookmarks are big part of my personal knowledge management system using Obsidian. For long time, I bookmarked articles to be read in later time. Now I’m looking for a way to generate summary for such bookmark notes inside Obsidian. Is there a plugin that can do this directly inside Obsidian without going out for another websites to do the summarization?
Normally external link/url to the article is stored in frontmatter “url”, so perhaps this AI plugin can accept this as a variable for example as: {{url}}
WebClipper Extension
You could use the Obsidian Webclipper extension for your browser. I haven’t extensively read through the documentation for the extension, however, I do know for a fact that you have the ability to set it up so that you can integrate LLM (Local or API) to provide something like this based upon the page you are clipping.
The wild thing is that you can can actually get really specific with what you’re targetting.
You can keep it simple or get really technical, but the point is that… I think you can do what you’re aiming for, before the note is imported to Obsidian via the webclipper extension.
Inside Obsidian
In Obsidian you could run either a small local model or the API to generate that summary for you with a saved prompt. This is probably less effort that learning the more technical stuff about the webclipper extension. All you’d need to do is:
-
Install Ollama on your computer
-
Pick a model from the Ollama model list
-
Open your terminal and download it by running ollama pull [your model here]
- I’d pick something like gemma 3:4b
-
Opening Obsidian and getting a plugin like text generator
Note:
If you prefer OpenAI, Anthropic or other non-local options you can skip the steps 1-3 and jump straight into 4, but you’ll need to have your API-Key from your paid subscription ready.
I Recommended the text generator plugin because it is pretty easy to save a template and run the model after the setup.
Hope this helps.
1 Like
Thanks, @Kinder, “Text Generator” plugin seems to be quite interesting plugin. I’ll try it.