spam-classifier

Simple LLM-based spam classifier for form submissions.

Install

npm install spam-classifier

Usage

import { createClassifier } from "spam-classifier";

// Set up once
const isSpam = createClassifier({
  apiKey: process.env.OPENROUTER_API_KEY!,
});

// Use anywhere - returns true/false
if (await isSpam("Need a quote for your services")) {
  // spam - ignore it
}

// Or get details
const result = await isSpam.details("Need a quote");
// { isSpam: false, confidence: 0.95, reason: "Customer requesting quote" }

Options

| Option | Type | Default | Description | |--------|------|---------|-------------| | apiKey | string | required | API key | | model | string | "openai/gpt-4o-mini" | Model to use | | baseUrl | string | OpenRouter | API endpoint | | zdr | boolean | false | Enable Zero Data Retention |

Providers

Works with any OpenAI-compatible API:

// OpenRouter (default)
const isSpam = createClassifier({ apiKey: process.env.OPENROUTER_API_KEY! });

// OpenAI directly
const isSpam = createClassifier({
  apiKey: process.env.OPENAI_API_KEY!,
  baseUrl: "https://api.openai.com/v1/chat/completions",
  model: "gpt-4o-mini",
});

// Local LLM
const isSpam = createClassifier({
  apiKey: "not-needed",
  baseUrl: "http://localhost:11434/v1/chat/completions",
  model: "llama3",
});

Fail-Safe

Returns false (legitimate) if the API fails, so you never block real customers.

Cost

~$0.0002 per classification. 10,000 submissions = ~$2.

License

MIT