Spam check names

Every email we receive on a custom domain goes through a series of spam checks. Each check that fires adds (or subtracts) points, and the sum is the spam score. The checks themselves are listed by name in your webhook payload, under spam.symbols, and you can write alias rules on those names directly.

Note: Spam analysis and alias rules are Maker+ features.

Where to find the names

Every scored payload carries the checks that fired, with what each one added:

"spam": {
  "score": 9.2,
  "symbols": [
    { "name": "FROM_BRAND_LOOKALIKE", "weight": 5, "description": "kraken.com.example-attacker.nl" },
    { "name": "FROM_LOOKALIKE_DOMAIN", "weight": 3, "description": "kraken.com.example-attacker.nl" },
    { "name": "TO_DN_RECIPIENTS", "weight": 2, "description": null }
  ]
}
  • name is what you put in a rule. Copy it exactly from a payload rather than guessing.
  • weight is what the check added to the score. A check can fire with a weight of 0: it is listed, adds nothing, and still counts in a rule.
  • description is the detail the check reported, such as the domain it looked at, or null.

To see which checks mark the mail you want to catch, open that email in your logs and view its webhook payload, or look at what your webhook received.

Acting on a check: "Spam rule fired"

In the alias rules editor, the Spam rule fired condition matches when any of the checks you name fired on the message:

Action:    Block
Condition: Spam rule fired  is one of  FROM_BRAND_LOOKALIKE, PHISHED_OPENPHISH

It matches whatever the total score. That is the point of it: a phishing email can score low overall because it passes sender authentication, while one check identifies exactly what it is.

  • Whole names, any case. phished_openphish matches PHISHED_OPENPHISH; PHISH matches nothing.
  • Several names, commas. The condition matches if any one of them fired.
  • Only names are accepted (letters, digits and underscores). A misspelt name matches nothing, and under an Allow rule that would reject everything, so the editor refuses anything that is not a name.
  • Unscored mail is not decided by it. If the spam scanner did not run for a message, the condition is undecidable; when the outcome depends on it, the email is delivered, not blocked. A message that was scored but on which none of your checks fired is an ordinary "no match".

Not sure yet? Use Tag instead of Block: matching mail is delivered with an X-EmailConnect-Tagged: spam header, so you can see what the condition would catch.

Common checks

The scanner has many checks; these are the ones most useful in a rule. All of them appear by these exact names in spam.symbols.

Name Fires when
FROM_LOOKALIKE_DOMAIN The sender's domain has the shape <name>.com.<domain>.<tld>, for example support@brand.com.example-attacker.nl: a registrable-looking name used as part of someone else's domain. Adds 3
FROM_BRAND_LOOKALIKE The same shape, with a known brand (a bank, a payment or crypto service, a big tech company) in front. Adds 5, on top of FROM_LOOKALIKE_DOMAIN
PHISHING A link's visible text names a different site than the one it leads to
PHISHED_OPENPHISH A link points to a page listed in the OpenPhish feed of known phishing sites
DBL_SPAM A domain in the message is on Spamhaus's list of spam domains
R_SPF_FAIL The sending server is not allowed to send for the sender's domain (SPF). Common on forwarded mail, so on its own it proves little
VIOLATED_DIRECT_SPF The SPF check failed on mail that was not forwarded
DMARC_POLICY_QUARANTINE, DMARC_POLICY_REJECT The message failed DMARC, and the sender's domain asks receivers to quarantine or reject such mail

Lookalike domains deserve a word: they pass sender authentication (SPF, DKIM, DMARC), because the attacker set those up properly for their own domain. Authentication proves who sent a message, not that it is from the brand it names. That is why the lookalike checks look at the shape of the domain instead.

Recipes

Block lookalike-domain phishing, whatever it scores

Action:      Block
Condition 1: Spam rule fired  is one of  FROM_BRAND_LOOKALIKE, FROM_LOOKALIKE_DOMAIN

Block known phishing links, and spam above a threshold

Action:      Block
Condition 1: Spam rule fired  is one of  PHISHED_OPENPHISH, PHISHING
Condition 2: OR  Spam score  greater than  15

When a name stops matching

The check names come from the spam scanner. A scanner upgrade can occasionally rename a check, after which a condition naming the old name no longer matches. A Tag rule on the same names makes that easy to notice: tagged mail stops arriving.

Related topics