AI agent for community moderators
Ban Appeal Review Agent
Fair, consistent appeal decisions with clear reasons
What it does
Ban appeals take time, and inconsistent decisions erode trust in the moderation team. When an appeal comes in, this agent gathers the original decision, the content involved, the user's history, the guideline cited and the appeal text. It checks whether the original action followed the guidelines and the penalty ladder. It also checks whether the content was misread, such as quoted text or sarcasm between friends. It drafts a recommendation to uphold, reduce or reverse, with reasons. Then it checks the recommendation against past appeals. If it is inconsistent, it explains why or changes the recommendation. A different moderator from the one who issued the ban makes the decision. Edge case: appeals for safety bans, like threats, always need two moderators to agree.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Appeal filed
- Gather decision, content, history and guideline
- Check action against guideline and penalty ladder
- Check for misread context
- Draft recommendation and response
- Is the recommendation consistent with past appeals?If not: explain the difference or adjust the recommendation. Back to step 3.
- A different moderator decides and approves the replyThe agent waits here for your OK.
- Appeal closed and logged
How it decides
It compares the action with the penalty ladder and past appeals and recommends the outcome most consistent with both.
- The original moderator never decides the appeal
- Safety bans need two moderators
- Penalty must match the ladder step
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Penalty ladder
- Response time target
- Which bans need two reviewers
- Reply template tone
What keeps you in control
It always asks you first
- Appeal decision
- Sending the reply
Hard limits
- Never reverses bans itself
- Never shares reporter identities
It stops when
- Done: appeal closed
- Stop: legal request involved, hand to admin
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide
