antiscamai

AI-powered request inspection middleware for Python.
Add one line to your FastAPI, Django, or Flask app and every incoming request is automatically screened by a trained AI model for scams, phishing, social engineering, and fraudulent content.

Install

# FastAPI / Starlette
pip install antiscamai[fastapi]

# Django
pip install antiscamai[django]

# Flask
pip install antiscamai[flask]

# All frameworks
pip install antiscamai[all]

Quick Start

FastAPI (Middleware)

from fastapi import FastAPI
from antiscamai.middleware.fastapi import AntiScamFastAPIMiddleware
import os

app = FastAPI()
app.add_middleware(
    AntiScamFastAPIMiddleware,
    api_key=os.environ["ANTISCAM_API_KEY"],
    mode="block",   # "block" | "flag" | "monitor"
)

# All your existing routes are now protected — no other changes needed
@app.post("/api/contact")
async def contact(body: ContactForm):
    ...

FastAPI (Per-Route Dependency)

from fastapi import Depends
from antiscamai.middleware.fastapi import antiscam_fastapi

checker = antiscam_fastapi(api_key=os.environ["ANTISCAM_API_KEY"])

@app.post("/api/payment")
async def payment(body: PaymentForm, _=Depends(checker)):
    ...

Django

# settings.py
ANTISCAMAI = {
    "API_KEY": os.environ.get("ANTISCAM_API_KEY"),
    "MODE": "block",                              # block | flag | monitor
    "ENDPOINT": "http://localhost:5000",
    "EXCLUDE_PATHS": ["/health/", "/static/"],
    "INSPECT_METHODS": ["POST", "PUT", "PATCH"],
}

MIDDLEWARE = [
    # ... your existing middleware ...
    "antiscamai.middleware.django.AntiScamDjangoMiddleware",
]

Flask

from flask import Flask
from antiscamai.middleware.flask import antiscam_flask
import os

app = Flask(__name__)
antiscam_flask(app, api_key=os.environ["ANTISCAM_API_KEY"])

What Gets Blocked

Response on Blocked Request

HTTP 403 Forbidden
{
  "error": "Request blocked by AntiScam AI",
  "requestId": "a4f2c1d3...",
  "riskLevel": "HIGH",
  "reason": "High urgency language + suspicious patterns detected"
}

Read the Inspection Result

# FastAPI — result stored in request.state
@app.post("/api/contact")
async def contact(request: Request, body: ContactForm):
    result = request.state.antiscam
    # result.threat_score  → 0–100
    # result.risk_level    → MINIMAL | LOW | MEDIUM | HIGH | CRITICAL
    # result.decision      → allow | flag | block
    # result.threats       → list of detected threats with explanations

# Django — result stored in request.antiscam
def my_view(request):
    result = getattr(request, "antiscam", None)

# Flask — result stored in g.antiscam
from flask import g
result = g.antiscam

Threat Callback

from antiscamai.middleware.fastapi import AntiScamFastAPIMiddleware

async def on_threat(result):
    print(f"Threat! score={result.threat_score} level={result.risk_level}")
    # Send to Slack, PagerDuty, your SIEM...

app.add_middleware(
    AntiScamFastAPIMiddleware,
    api_key="...",
    on_threat=on_threat,
)

Modes

Configuration

Self-Hosting

git clone https://github.com/antiscamai/backend
cd backend/deploy/docker
docker-compose up -d

AntiScamFastAPIMiddleware(app, api_key="YOUR_KEY", endpoint="http://localhost:5000")

Requirements

• Python 3.10+
• httpx >= 0.27.0

License

MIT © AntiScam AI