Skip to main content

✨ Smart Triggers

Smart Triggers are moderation rules written in plain English. Instead of matching exact keywords, you describe what kind of message should trigger an action — the AI evaluates each message against your prompt and acts on confident matches.

Overview​

When configured, the agent can:

  • Judge every message against prompts you write, with no keyword lists to maintain
  • Apply each rule to all members or only to new members
  • Delete, warn, timeout, kick, or ban on a match — or respond with a reply or a channel message
  • Catch things keyword filters can't express, like intent, tone, or impersonation attempts

Configuration​

To configure Smart Triggers:

  1. Open your project in the Fren One webapp.
  2. Select the Discord server you want to configure.
  3. In the sidebar under Agent Configuration, select AI Features, then the Smart Triggers tab.
  4. Toggle "Let AI evaluate messages against custom prompts" on.
  5. Select Add Trigger and fill in the editor.

Each trigger in the list shows its prompt, action pills, a "New members only" pill when scoped, and a plain-English summary of what it does.

Prompt​

Under "When should this trigger fire?", describe what kind of message should trigger an action — the AI evaluates each message against this prompt. Required. Write it as a yes/no decision the AI can make about a single message:

Example prompts
  • "Is this message asking for the official contract address?"
  • "Is this message impersonating support staff?"
  • "Does this message contain hate speech or a personal attack?"

Scope​

Under "Who does this apply to?", choose All Members or New Members Only. New members are defined by the new-member criteria on the Moderation page.

Actions​

Under "What should happen?", select at least one action:

ActionDescription
Delete messageRemoves the matching message.
Warn memberIssues a warning through the warning system.
Timeout memberTimes the member out. Reveals Timeout duration (seconds) — up to 2,419,200 seconds (28 days).
Kick memberKicks the member from the server.
Ban memberBans the member.
Reply to messageReplies to the matching message. Reveals a required Reply message field.
Send messagePosts a message to the channel. Reveals a required Message field.

Some combinations are mutually exclusive:

  • Only one punishment at a time — Warn, Timeout, Kick, and Ban can't be combined.
  • Reply to message and Send message are exclusive — pick one.
  • Delete message and Reply to message are incompatible (there's nothing left to reply to).

How it works in Discord​

The AI evaluates each new message against your prompts as a background task, so actions fire with a slight delay after the message is posted. Only high-confidence matches (0.85 or above) execute the configured actions — borderline messages are left alone.

Smart Triggers vs. Triggers vs. Automations​

Three systems act on messages; pick the right one:

  • Smart Triggers (this page) — the AI interprets a natural-language prompt. Best for rules about meaning or intent.
  • Keyword Triggers — exact pattern matching (contains, exact, starts with, ends with, regex). Instant, deterministic, no AI required.
  • Automations — event → condition → action workflows that go beyond messages (reactions, role changes, threads, voice, joins).

Saving your changes​

This page uses a batched save — new, edited, and deleted triggers stay local until you select Save in the bar at the bottom of the page. Deleting a trigger marks it as pending with an undo option until you save; leaving with unsaved changes shows a warning.

Best practices​

Writing prompts that work
  • Phrase each prompt as a yes/no question about a single message — the AI answers it for every message it sees.
  • One behavior per trigger. Two narrow triggers beat one broad, ambiguous one.
  • Start with Reply to message or Warn member to see what a new prompt matches before escalating to kicks or bans.
  • Use New Members Only scope for scam patterns that established members never post — it cuts false positives on regulars.
AI judgment on punishments

A smart trigger with Ban member or Kick member acts on the AI's judgment, not an exact match. Test new prompts with a milder action first.

Troubleshooting

A trigger isn't firing?

  • Check the master toggle "Let AI evaluate messages against custom prompts" is on and your changes are saved.
  • The AI only acts on high-confidence matches — a vague prompt produces uncertain judgments that don't fire. Make the prompt more specific.
  • Smart triggers require the AI service; they no-op silently when it's unavailable.

Actions seem delayed?

  • Evaluation runs in the background, so a short delay between the message and the action is normal.

On this page