Ban lists and the whitelist

Lists are the instant tier of screening: every incoming message is checked against your subscribed lists before any AI runs. A match flags immediately with the matched term highlighted.

System lists and your lists

• System lists are curated by Observer and read-only — subscribe or unsubscribe per list under Moderation → Lists.
• Org lists are yours: create one, add terms, delete terms. Each entry carries its own severity, which feeds the severity → action policy.

Importing an existing list

The entries editor takes bulk paste: one term per line, # for comments, one severity applied to the batch. Duplicates are merged silently, so re-importing an updated list is safe.

Matching is normalization-aware

Terms match against a normalized form of the message — case, common leet substitutions, and spacing tricks don't dodge a list entry. The queue row shows both the raw text and the normalized match so you can see why it hit.

The actor whitelist

Org-wide, per-actor: accounts whose messages skip screening entirely — announcement bots, bridge bots, service accounts. Whitelisted actors' messages still land in your logs; they're exempt from screening, not from the record. Each entry records who added it and why.

For Discord, role-based bypass is usually the better tool — see trusted roles. The whitelist is for specific accounts on any platform.