kick

Delete ALL messages from the bot in the current channel.

purge

refresh_rules

Open an interactive panel to configure Modcord for this server.

settings

Show current per-guild settings as raw JSON. For debugging purposes.

settings-dump

show_rules

Checks if the bot is online and its round trip time.

test

timeout

warn

# Modcord – AI-Powered Discord Moderation

> **Stop moderating like it's 2015.** Modcord understands *context*.

## What is Modcord?

Modcord is an **AI-driven moderation bot** that protects your Discord server using large language models to detect toxicity, spam, raids, and harmful behavior—not just keyword matching.

Unlike traditional bots that ban memes for containing "bad words," Modcord reads the conversation. It understands sarcasm, context, and intent. It knows the difference between a real threat and banter between friends.

---

## Core Features

### **Context-Aware AI Detection**
- Analyzes *complete conversations*, not isolated messages
- Detects hate speech, harassment, and toxic behavior with near-human accuracy
- Works with any LLM-compatible API (OpenAI, Anthropic, local models, etc.)

### **Behavioral Analysis**
- Tracks member activity patterns to flag coordinated raids in real-time
- Identifies self-bots and malicious actors before they cause damage
- Learns from your moderation decisions to improve over time

### **Fully Customizable**
- Define server-specific rules and let the AI learn them
- Set channel-specific guidelines (off-topic content in #general vs #random)
- Exempt users and roles from automated moderation
- Control which actions auto-trigger (warn, delete, timeout, kick, ban)

### **Transparent Moderation**
- Every action comes with a **reason** your mods can read
- See exactly why the bot made a decision
- Appeal system-friendly (no black-box bans)
- Full audit logging for compliance

### **Built for Developers**
- **100% open source** – Audit the code yourself
- **Self-hosted** – Your data never leaves your server
- **Java/Gradle** – Production-grade, battle-tested stack
- **Database-backed** – PostgreSQL for persistence and scale

---

## How It Works

1. **Messages arrive** → Bot batches them with recent history
2. **Context built** → Pulls older messages for conversation context
3. **AI analyzes** → Sends structured data to your LLM endpoint
4. **Decision made** → Gets back moderation action (warn, delete, timeout, etc.)
5. **Action applied** → Deletes messages, logs to audit channel, DMs user

All configurable. All transparent.

---


## Perfect For:

✅ **Gaming communities** – Catch toxicity before it escalates  
✅ **Study/work servers** – Keep discussions professional  
✅ **Creator communities** – Protect members from harassment  
✅ **Large servers** – Handle scale without manual review  
✅ **Privacy-conscious admins** – Self-host, stay in control  

---

## Getting Started

### Quick Invite
[Add Modcord to your server](https://discord.com/oauth2/authorize?client_id=YOUR_BOT_ID)

### Commands
- `/status health` – Check bot health
- `/exclude add @user` – Exempt user from moderation
- `/mod warn @user [reason]` – Manual warning
- `/debug show-rules` – View current server rules

### Documentation
Full setup guide and architecture docs available on **[GitHub](https://github.com/HoneyBerries/Modcord)**

---

## Privacy & Security

- ✅ Open source – Review every line of code
- ✅ Self-hosted option – Run on your own infrastructure
- ✅ No third-party tracking – Only connects to your LLM provider
- ✅ Database-backed audit log – Full compliance trail
- ✅ GDPR-friendly – Built to respect user data

---

## Support

- **Discord Server:** [Join our community](https://discord.gg/tSMgBPYb48)
- **GitHub Issues:** Report bugs and request features
- **Source Code:** [github.com/HoneyBerries/Modcord](https://github.com/HoneyBerries/Modcord)

---

## 🎯 Why Modcord?

| Feature | Modcord | Keyword Filters | Manual Review |
|---------|---------|-----------------|---------------|
| **Understands context** | ✅ | ❌ | ✅ |
| **Scales to 1000+ members** | ✅ | ✅ | ❌ |
| **24/7 automatic** | ✅ | ✅ | ❌ |
| **Explainable decisions** | ✅ | ❌ | ✅ |
| **Open source** | ✅ | ❌ | N/A |
| **Customizable to *your* rules** | ✅ | ⚠️ Limited | ✅ |

---

## Status

**Active Development** – Latest updates at [github.com/HoneyBerries/Modcord](https://github.com/HoneyBerries/Modcord)

---

*Modcord: Moderation that thinks.*

Modcord

Context-aware AI moderation: Detects toxicity, raids & spam. LLM-powered, fully customizable, and open source.

Vote cho Modcord

Bạn muốn hỗ trợ chúng tôi?