Troubleshooting

If the bot is misbehaving, 90% of the time it's one of the issues below. Work through them top-to-bottom before opening a support ticket.

#"The bot isn't doing anything"

Is it enabled? Run /modconfig. If status is Disabled, run /enable.
Is dry-run on? /modconfig will show 🧪 if so. Dry-run logs decisions but takes no action — run /dryrun mode:off.
Is the channel in the moderated list? Only channels you set with /setmodchannels are continuously moderated. Other channels only respond to @mention scans.
Can the bot read the channel? Server Settings → Roles → Blify Mod → Permissions, plus the channel-specific overrides. It needs at least View Channel, Read Message History, and Send Messages.

The log embed will show Failed with a reason. The usual suspects:

"Missing Permissions" — the bot's role is at or below the offender's highest role. Move the bot's role up in Server Settings → Roles.
The target is the server owner. Discord will not allow this regardless of permissions or role position.
The user is immune. Check /immuneuser list and /immunerole list.
Bot is missing a permission. It needs Manage Messages (delete), Moderate Members (timeout), Kick Members, Ban Members. Re-invite with the default permission set if any are missing.

This shouldn't happen — the system prompt explicitly forbids enforcing un-listed rules. If you see it:

Check /setrules list. Is there a rule that technically covers it, even if you didn't intend it to?
If yes, edit your rules: /setrules remove number:N and re-add a tighter version.
If no, please report it in the support server with the log embed screenshot — that's a model error and we want to see it.

Too harsh: lower the strictness cap with /setstrictness level:medium (or low).
Too lenient: make sure your rules are specific. A rule titled "Don't be mean" is too vague — Claude will only flag the most obvious cases. "No targeted insults or personal attacks against other members" is much sharper.
Recurring miscalls: consider Pro/Premium. Claude Sonnet 4.6 is meaningfully better on subtle calls than the free tier's gpt-5-mini judge.

If OpenAI or Anthropic is having an outage, the screen step has automatic fallback and retries — this can add a few seconds.
During peak hours the bot caps concurrent AI requests to avoid rate-limit bursts; at extremely high traffic a request can queue for a moment.
If it consistently takes >30s, check the status page or the support server.

The bot does three quick checks before saving a new rule:

Duplicate title — pick a different title or remove the existing one first.
Multi-rule submission — the rule bundles two unrelated topics (e.g. "no politics AND no spam"). Split it into two separate rules. There's a 30-second cooldown after a multi-rule rejection to prevent spam.
Policy violation — the rule would force the bot to itself violate Discord's Community Guidelines (e.g. a rule requiring it to promote hate speech). Rules banning hate speech are fine; rules requiring it are not. The error message will explain.

Make sure you're in the support server — the Pro/Premium role lives there.
Patreon role-grants can take a few minutes after pledging. If it's been more than an hour, open a ticket.
After the role appears, run /pro claim in the server you want the perks on. Pro doesn't auto-apply — you choose which servers use your slot(s).

Top.gg's webhooks occasionally lag. Run /credits after a few minutes — if it shows the new credits, all good.
If you hit "vote" but never actually completed the captcha, top.gg counts that as no vote.
The bot deduplicates within a 1-hour window and a UTC-day window — repeating a vote within either won't double-credit you. (One vote per 12-hour cooldown anyway.)

Still stuck? Support server — include a screenshot of /modconfig and the relevant log embed.