Could a Core Node Add-On Store Work?

I've been thinking about creating a store for add-ons that core node users could use. Basically, it would be a site or app to add filters to their core nodes. The goal is to tackle the issue of mempool policy centralization. For years, the core team hasn’t given us the filters we need, rejecting ideas like the inscriptions filter proposed by Luke Dashjr. They want 99% of the network to stick to their choices.

But how can a third-party app or site actually change how Bitcoin Core nodes connect? If it were that easy, we'd have a bunch of apps with different features, but what about security? Some risks might have been fixed before, and letting users run anything could open up a can of worms.

Adding to that, when you mess with node software, especially Bitcoin's, it’s like opening Pandora's box. There’s a consensus thing that makes it complicated. Upgrade pointed out security risks with a core app store. It's hard to get everyone on board when those risks are involved. Plus, cypherpunk folks generally distrust third-party apps, it’s like browser extensions to them.

So are you saying it’s like downloading a browser extension from Chrome or Firefox? I just can’t see how anyone would want to deal with an add-on like that.

This is never gonna happen. Who in their right mind would want third-party stuff on their node, especially if they’ve got BTC in it? What if some bad settings mess things up? Are you gonna take the blame if someone loses money? Everything you need can be changed in the config file, and you can do that manually.

If you go that route, block propagation could get worse. If miners start producing completely different blocks, it’ll create chaos. Everyone would be downloading a ton of data, and if those transactions aren’t relayed in the network, they’ll just go outside it. That’s not good.

Here’s how I see it: every add-on would go on GitHub, open source for anyone to check out. I think we could implement this in a few ways. One, edit the core code and recompile, but that’s super inefficient. Or two, don’t touch the core at all and intercept transactions at the node level. Like, if you want to filter large op_returns, we can do that before they even hit the node.