Discord moderation with a log you can read
Every action is recorded with who took it, against whom, and why.
Moderators cannot act upwards
The hierarchy is checked on the server, not in the interface: a moderator cannot time out an administrator, and an administrator cannot ban the owner. Hiding the button is not the same as refusing the action, and only one of the two survives somebody calling the API directly.
Warnings that expire
A warning from four months ago should not weigh the same as one from yesterday. Warnings age out on a schedule you set, so somebody who had a bad week in spring is not carrying it in autumn, and the moderator sees the current picture rather than a lifetime total.
The spam filter stores no message text
Duplicate detection asks one question: is this the same message again? That is an equality test, and a keyed one-way fingerprint answers it exactly as well as the text does. The fingerprints live for seconds and cannot be turned back into words. The column that used to hold text is gone, and the published schema shows it.
On the free tier
The free tier keeps 7 days of audit history; Premium keeps 365.
Premium is $9 a month. The free tier needs no payment and is not a trial.
Questions
- Does it read my members' messages?
- It counts that a message happened and, for the spam filter, keeps a keyed fingerprint of the last few for a few seconds. It never stores the text. There is a test that fails if raw text appears in the database.
- Can I see what a moderator did last month?
- As far back as your plan's window: 7 days on Free, 365 on Premium. Older entries are removed by a nightly job, and the logs page says when it last ran.