Modstr

Smart Moderation for the Nostr Protocol.

Modstr is a smart moderation service for the Nostr protocol. It analyzes, classifies, and labels Nostr events—including text notes and media—using LLMs. Modstr can be integrated with both relays and clients via HTTP and WebSocket interfaces, enabling automated moderation pipelines or in-app content safety features.

✨ Features

• ✅ Event Classification Engine Categorize events as SAFE, WARNING, or HARMFUL with detailed labels like SPAM, INFORMATIVE, OFFENSIVE, and more.

• 🤖 AI-Powered Moderation Use LLMs or custom rules to assess content safety, language, sentiment, or compliance.

• 🌐 Relay + Client Friendly Works with relays for moderation pipelines, or clients for in-app safety.

🛠️ Installation

git clone https://github.com/ZigBalthazar/mostr.git
cd mostr
npm install

🚀 Usage

Note
📘 Full API documentation is available on Dezh Docs.

To analyze a Nostr event, simply make a POST request to the /analyze endpoint with a valid event payload:

Example

curl -X POST 'http://{BASE_URL}/analyze' \
  -H 'Accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
    "id": "259bf57751b9d75da8d99eda9b3fb842a58c7310bc097e408ddbe1a2fdd29a2c",
    "sig": "60fcfecb23c779ef991415e976ebeaae47a8baaa945c70fb3b1672d82f5c2ab249b97c18756780f951bb17224a58589c27539fffbd117302b1eccf45afdf5807",
    "kind": 1,
    "tags": [],
    "pubkey": "bd4ae3e67e29964d494172261dc45395c89f6bd2e774642e366127171dfb81f5",
    "content": "It's getting complicated.",
    "created_at": 1748264568
  }'

Sample Response

{
  "success": true,
  "message": "Classification generated successfully.",
  "responseObject": {
    "event_id": "259bf57751b9d75da8d99eda9b3fb842a58c7310bc097e408ddbe1a2fdd29a2c",
    "status": "SAFE",
    "labels": ["OK", "INFORMATIVE"],
    "reason": "The text is a simple statement expressing a feeling or observation. It doesn't contain any harmful content."
  },
  "statusCode": 200
}

---

🌉 Integrating with Relays

Strfry Integration

To integrate Modstr with relays using Strfry:

1. Use the Strfry plugin(ts) available in the /plugin directory.
2. Refer to the Strfry plugin documentation for setup instructions.

📖 Labels & Statuses

Statuses:

• SAFE
• WARNING
• HARMFUL

Labels:

[
  "OK", "INFORMATIVE", "FRIENDLY", "OFFENSIVE", "SPAM", "MISLEADING",
  "SENSITIVE", "PROFANITY", "HATEFUL", "VIOLENT", "SEXUAL", "SELF_HARM",
  "DANGEROUS", "NSFW", "AI_GENERATED", "UNVERIFIED", "LANGUAGE_UNKNOWN"
]

📚 Roadmap

• Full moderation with integration to a web of trust service
• Local model support (e.g., Ollama, Llama.cpp)
• Websocket interface integration
• Language detection integration
• NIP-90 data vending machine integration
• Event suggestion service (based on user following and liked events)
• Complete documentation on how to run, use, and integrate with Modstr

🤝 Contributing

Contributions are welcome! Please open issues or pull requests.

📄 License

MIT