Ivory

Ivory is an automoderator and anti-spam tool for Mastodon admins. It automates handling certain trivial reports and new user requests using rules - configurable tests that check reports and pending users for bad usernames, malicious links, and more.

Installation Guide

This installation guide assumes you know your way around a Linux terminal, have Python and Git installed, and maybe a little bit about common tech like Python and JSON formatting.

Installing

In a Linux terminal, the following commands will clone and install Ivory to whichever folder you're in. Make sure you have Git and Python installed:

git clone https://github.com/bclindner/ivory
cd ivory
python -m venv .
source bin/activate
python -m pip install -r requirements.txt

This repo also comes with a Dockerfile, so if you want to deploy with that, that works too:

git clone https://github.com/bclindner/ivory
cd ivory
docker build -t ivory .
docker run -v /srv/ivory/ivory_config.json:/app/config.json ivory

Configuration

Before starting Ivory, you need to create a new application in the Preferences menu. Don't worry about setting redirect URIs or anything that isn't required - just make sure you enable all of the admin: scopes. Once you've created the application, you'll want to grab its access token to place in the configuration file (example below).

Be EXTREMELY careful with handling the access token this generates - this key has a lot of power and in the wrong hands, this could mean someone completely obliterating your instance.

Once you've done that, it's time to set up your config file. Configuring Ivory is done with JSON; a sample is below:

{
  "token": "<YOUR_ACCESS_TOKEN_HERE>",
  "instanceURL": "<YOUR_INSTANCE_URL_HERE>",
  "waitTime": 300,
  "reports": {
    "rules": [
      {
        "name": "No spam links",
        "type": "link_resolver",
        "blocked": ["evilspam\\.website", "dontmarry\\.com"],
        "severity": 2,
        "punishment": {
          "type": "suspend",
          "message": "Your account has been suspended for spamming."
        }
      },
      {
        "name": "No link-in-bio spammers",
        "type": "bio_content",
        "blocked": ["sexie.ru"],
        "severity": 1,
        "punishment": {
          "type": "disable",
          "message": "Your account has been disabled for spamming."
        }
      }
    ]
  },
  "pendingAccounts": {
    "rules": [
      {
        "name": "No <a> tags",
        "_comment": "Because honestly, you're definitely a bot if you're putting <a> tags into the field",
        "type": "message_content",
        "blocked": ["<a href=\".*\">.*</a>"],
        "severity": 1,
        "punishment": {
          "type": "reject"
        }
      },
      {
        "name": "StopForumSpam test",
        "type": "stopforumspam",
        "threshold": 95,
        "severity": 1,
        "punishment": {
          "type": "reject"
        }
      }
    ]
  }
}

A more in-depth guide to Ivory configuration and the list of rules and punishments can be found on the wiki.

Ideally you only have to change this once in a blue moon, but if you do, you can use the "dryRun": true option to prevent Ivory from taking action, so you can test some rules on recent live moderation queues.

Running

After you've set up a config file, run the following in a Linux terminal:

# if you're running in the same terminal session you installed from, you can
# skip this next line:
source bin/activate
python .

Hopefully, no errors will be thrown and Ivory will start up and begin its first moderation pass, reading the first page of active reports and pending users and applying your set rules. Ivory will handle these queues every 300 seconds, or 5 minutes. (This is controlled by the waitTime part of the above config file - if you wanted 10 minutes, you could set it to 600!)

If you'd rather run it on some other schedule via a proper task scheduler like cron or a systemd .timer unit, you can use python . oneshot which will run Ivory only once. This sample cron line will run Ivory every 5 minutes and output to a log file:

*/5 * * * * cd /absolute/path/to/ivory; ./bin/python . oneshot >> ivory.log

Extending (custom rules)

You'll notice the rules/ folder is a flat folder of Python scripts, one per Ivory rule. If you've got a little Python experience, you can easily create your own rules by just dropping in a new Python file and using one of the other files in the folder as a jumping-off point.

The reports and pending accounts that Ivory rules receive are the same as what Mastodon.py provides for reports and admin accounts, respectively.

Don't forget to use dryRun in your config when testing your new rule!

Once you've finished writing up your custom rule, say as rules/filename_of_your_rule.py, you can address it by its filename in your config:

...
"pendingAccounts": {
  "rules": [
    {
      "name": "An instance of my cool new rule",
      "type": "filename_of_your_rule",
      "custom_option": true,
      "severity": 5,
      "punishment": {
        "type": "reject"
      }
    },
  ]
}
...

If you come up with any useful rules and wouldn't mind writing a schema and some tests for it, making a pull request to include it in Ivory's main release would be highly appreciated! The more rules Ivory gets, the more tools are collectively available to other admins for dealing with spammers and other threats to the Fediverse at large.

Bugs & Contributing

If you have any issues with Ivory not working as expected, please file a GitHub issue.

Contributions are also welcome - send in pull requests if you have anything new to add.

If you have any other questions, go ahead and ping me on Mastodon and I might be able to answer them.