The AI boom has resulted an absolute explosion of plugins. I used to follow along with new ones, when there were just a few dozen, then a few hundred, then a couple thousand. Now it feels like chaos. It takes 9 entire seconds just to change the sort order on the community plugins panel on my reasonably powerful Mac Mini M4 Pro.
It feels increasingly unsafe. How can Obsidian possibly even begin to audit any of these for malicious code? I’m feeling depressed about it and wanted to share, and see if anyone has any ideas. Am I overreacting? I don’t want to switch to restricted mode, but this unbounded growth is also not comfortable.
I wish this was handled similar to the way e.g. Alpine package repos are: a main or core repo with a small amount of extremely well tested and trusted plugins, and then a community or edge repo that could be enabled where it’s a free-for-all and no guarantees of safety.
It’s arguably safer than it was before, because they now have automated scans for every update. That started in May: The future of Obsidian plugins - Obsidian (and the number of plugins has doubled since then — tho part of that is a big backlog that was released when they changed the review system). Before, each plugin only got one manual review when it was first released. You can see some details about the review results on the community site (which also has some helpful search tools that the in-app browser doesn’t yet): https://community.obsidian.md/.
I hear you about the overwhelm, tho. It’s a lot.
I thought something else must be going on but no, I get the same result on my M1. Wow!
I got overwhelmed by the whole plugin thing and left Obsidian for a long while because of it. Recently I decided to give it another whirl, and this time I’ve left it in restricted mode (i.e. turned off community plugins). This has let me focus more on thinking about how to use the existing functionality for what I need/want, which is quite extensive in its own right. So far I’m enjoying my return to Obsidian much more without the shadow of “plugin nervosa”.
Thanks. It’s somewhat comforting to know I’m not alone in feeling overwhelmed. There are some plugins I could probably learn to live without, but there are a core set that I absolutely would suffer without. So I’m not ready to go fully Restricted.
Out of the 33 plugins I currently have enabled, these 8 are critical to me- I don’t think I could live without them:
Advanced Tables - why this isn’t part of Obsidian is a mystery, tables are a mess without it
Advanced URI - littered all over my system are obsidian://adv-uri?… links so breaking all of those is not an option
Code Suite - being able to run codeblocks from my notes has become a must have
Custom Attachment Location - attachment folder would have been a complete mess and waste multiple GB of space without this
Paste Reformatter - I paste dozens of links a day, and manually removing the ?utm_* params and other fluff from them was wasting too much time
Recent Files - another “why isn’t this built in” feature
Templater - this one’s tough, but as of now I depend on a lot of the custom scripts that are run by this plugin
You are not overreacting. The number is actually growing faster than it looks. I checked community-plugins.json on September 21 and there were 7,882 plugins. Today there are 8,280. That is almost forty new plugins every single day.
What helped me stop feeling overwhelmed was realizing that the total number does not really matter. Download counts only tell you what used to be popular, not what is currently maintained. In the Obsidian ecosystem, these two things can be years apart. Here are some real examples from this month:
Calendar has over three million downloads, but its last release was a beta version from April 2021.
Full Calendar is archived, with its last release in April 2023.
Kanban literally says in its README that it is looking for new maintainers, and its last release was in May 2024.
Yet, people still recommend all three constantly. Because of this, before I install anything new, I spend two minutes checking its repository. I look at when the last release was, if the repo is archived, and if issues actually get answers. This simple check rules out a huge chunk of plugins you would otherwise have to test. Suddenly, those 8,280 plugins shrink to a much more manageable number.
Another thing that really helped me reduce my plugin count is the fact that Obsidian’s core features keep replacing them. It is worth reviewing your list once in a while. Bases can now handle a lot of the property querying you used to need Dataview for, and since version 1.14 it even has its own kanban board. The core app also got native table editing with drag handles a while ago, so it is worth checking if you really still need Advanced Tables. Every plugin you can replace with a core feature is one less third-party tool you have to keep trusting.
Your instinct to separate the 8 plugins you cannot live without from the other 25 is exactly right. Those 25 optional ones are where this kind of filtering pays off the most.
There have been bugs in all kinds of software all the time, but I’d be really surprised if many of the bugs I’ve noticed in the more recent past, wouldn’t be due to many devs/companys using AI in a sloppy way. I try to not use a lot of new software in general - but even apps I use since years, some more than a decade, started to introduce bugs in basic features that had no issues since forever
Wrt Obsidian - before I try/install any new plugin I usually visit the github page - when there are signs it’s made using AI I try to get a feeling for the experience of the dev. F.i. if I see someone having a hard time understanding basic bug reports or a simple featurerequest can’t be added (even while using AI) I skip the plugin. Chances are, the dev has not even remotely an idea what he/the AI is doing.
Sadly this is quite a bit of work and of course not a perfect check - I for sure will have both, false positives and true negatives, but I don’t have any better solution right now. What I can say is, that I’ve become way less interested in trying new plugins (or any other software).
I don’t bother installing community plugins, with the exception of the two authored by the Obsidian team (map view and importer).
I’ve found that most of the time community plugins don’t really assist with whatever I need. If I require a specific function in my workflow, I write my own plugin. then I get exactly what I want. If I think it will be useful to others, I’ll make the repo public on Github for others to discover.
For anyone who has even a remedial understanding of Node, you’ll be happier creating plugins speicifc to your workflow. No added bloat. No unnecessary functionality. Just exactly what you need.
— tho part of that is a big backlog that was released when they changed the review system). Before, each plugin only got one manual review when it was first released. You can see some details about the review results on the community site (which also has some helpful search tools that the in-app browser doesn’t yet): 