How to Set Your Flag and Block Thresholds

At the heart of Spamify is a score. Every submission accumulates points from the checks it fails — bad syntax, a dead mailbox, a tripped honeypot, an over-limit IP. Two settings turn that number into an action, and getting them right is the single most impactful tuning you’ll do.

The two thresholds

• flag_threshold — at this score, the submission is flagged for review. It still goes through, but it’s marked so you can inspect it.
• block_threshold — at this higher score, the submission is blocked outright.

Between them sits a gray zone: suspicious enough to flag, not damning enough to block. Below the flag threshold, submissions pass silently.

The mental model

score < flag_threshold        → allow (silent)
                    flag_threshold ≤ score < block_threshold → allow but flag for review
                    score ≥ block_threshold       → block

Set flag_threshold below block_threshold — the gap is your safety margin. A wide gap means more human review but fewer wrongly-blocked messages; a narrow gap means more automated blocking with less oversight.

How to tune, in practice

1. Start permissive. Begin with the defaults, which lean toward flagging over blocking.
2. Watch the dashboard for a week. Look at what landed in the flagged bucket. Is it clearly spam? Clearly real? Mixed?
3. If the flagged bucket is all obvious spam, lower block_threshold so those get blocked automatically.
4. If real messages are being flagged, raise flag_threshold a little, or allowlist the specific sender.

Thresholds and hard blocks

Some signals bypass the score entirely — the hard_block_mx and hard_block_smtp rules block an address the moment a mail server confirms it’s dead, regardless of threshold. Thresholds govern the accumulated score; hard blocks handle the certainties. And if you want the layer scores themselves adjusted, see tuning the engine weights.