NeuralTrust | Platform for Agent Security.

Middleware covers input. Guard completions in the route with direction="output" (Python SDK).

Coverage

Surface	Monitor	Block	Redact
LLM input	✅	⚠️	⚠️
LLM output	✅	⚠️	⚠️
Tool-level	⚠️	⚠️	⚠️

Ask — status is advisory: nothing prompts anyone unless your code does. A middleware has no user to prompt on the HTTP path either. Use it when you want every new AI route covered by configuration rather than by someone remembering to add the check. Not as your only defence when traffic reaches a model without an HTTP request — background jobs, queue consumers, internal calls all bypass it; those need the SDK at the call site. Limits. Use the async client in async frameworks, or every evaluation blocks the event loop. Scope the middleware to the AI routes so unrelated endpoints do not pay the latency. Output coverage needs a second evaluation on the response. Full comparison: Coverage.

import json

from fastapi import FastAPI, Request
from starlette.responses import JSONResponse
from trustguard import AsyncTrustGuard

app = FastAPI()
client = AsyncTrustGuard("<your-trustguard-url>", api_key="<collector-api-key>")

@app.middleware("http")
async def trustguard_middleware(request: Request, call_next):
    if request.method == "POST":
        body = await request.body()
        response = await client.guard(
            {"input": body.decode()},
            direction="input",
            consumer_id=request.headers.get("x-user-id", ""),
            session_id=request.cookies.get("session_id", ""),
        )
        if response.is_blocked:
            return JSONResponse({"detail": "Blocked by TrustGuard"}, status_code=403)
        if response.transformed_payload:
            request._body = json.dumps(response.transformed_payload).encode()
    return await call_next(request)

Introduction

Resources

Detector catalog

Integrations

Evaluate API

Python middleware

Coverage

​Coverage

Coverage