I added Dockerfiles to make containers of them for those running docker-compose Lemmy instances.
/0
Meta community. Discuss about this lemmy instance or lemmy in general.
Cheers mate, I'm in bed atm so I'll check tomorrow
I got the client to run in a container on a cronjob docker run --rm --name overseer-eye -e LEMMY_DOMAIN=discuss.online -e USERNAME=overseer -e PASSWORD=<PASSWORD> lemmy-overseer-eye:latest
I thought I'd containerize the server too, but it lacks a schema. Or I'm missing something?
The eeever is just a python script. Not sure what schema you need
The server has a database connection to sqlite or postgres.
It's setup to have one yes, but I haven't created a db yet. It's my next step
Well, that explains it then. Thanks. I thought I was going crazy. Sorry about that.
This is awesome, would you be open to contributions outside the realm of spam prevention?
I've been working on some scripts/apis to pre-search communities from federated instances so they immediately show up within an instance's search capability. Was also thinking about doing a bot account to auto subscribe.
Seems like this could fit with the whole overseer/curation theme. Would you agree or nah?
I'm always up for collaboration and prs. I'm adding now some cool new capabilities so once I have that you'll be able to build much easier on it
This sounds cool! Similar vein what I have been trying to do on Lemmyverse - Determine trust of federated instances based on trust lists from other instances, along with instance stats over time. I'm starting to collect stats like Instance User count by Time, so I can potentially export lists of instances with a given score for each Instance, which could be automatically read by your defederation scripts.
I like your threshold:
# If there's this many registered users per local post+comments, this site will be considered suspicious
ACTIVITY_SUSPICION = 20
I may steal some of your code to build out my scoring algo too :)
Feel free to Colab on the overseer directly too
I'll have a look through it tonight. 👌🔥🔥
Cool. I was wondering when someone would implement this sort of email RBL list.
I'm not finding any info on The Overseer how instances are deemed suspicious and what mechanisms there are for reporting/disputing. How are instances scored?