LLMAgentSecuritySaaS
Return to AIPromptShield Platform
APS Gateway Documentation

APS Gateway Integration Guide

APS Gateway is a zero-latency in-memory reverse proxy compatible with OpenAI, Anthropic, Google Gemini, and open-weight models. Protect applications from prompt injections, mask sensitive PII, and quarantine rogue agent actions in 2 lines of code.

The 2-Line Drop-In Replacement< 14ms Mean Pipeline Latency
Line 1: Target EndpointbaseURL = "https://gateway.aipromptshield.com/v1"

Reroutes OpenAI/Anthropic SDK calls to your nearest in-memory APS Edge cluster.

Line 2: Security Policyheaders = { "X-APS-Policy": "strict-defense" }

Activates heuristic classifiers, Presidio PII redaction, and vector anomaly isolation.

Python (OpenAI SDK v1.0+)
import os
from openai import OpenAI

# 2-Line Drop-In Proxy Integration:
client = OpenAI(
    api_key=os.environ["OPENAI_API_KEY"],
    base_url="https://gateway.aipromptshield.com/v1",  # <-- Line 1: Reroute to APS Edge Proxy
    default_headers={"X-APS-Policy": "strict-defense"} # <-- Line 2: Enforce zero-day guardrails
)

# Prompts & RAG contexts are scanned, PII masked, and validated in-flight (<40ms)
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Analyze customer database record..."}]
)
print(response.choices[0].message.content)
Node.js / TypeScript
import OpenAI from 'openai';

// 2-Line Drop-In Proxy Integration:
const client = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
  baseURL: 'https://gateway.aipromptshield.com/v1',    // <-- Line 1: APS Edge Proxy
  defaultHeaders: { 'X-APS-Policy': 'strict-defense' }, // <-- Line 2: Security Policy
});

// Outbound calls are protected with Presidio PII masking and agent sandboxing
const response = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Process customer refund request...' }],
});
console.log(response.choices[0].message.content);
LangChain & LangGraph
from langchain_openai import ChatOpenAI

# Drop-in LangChain / LangGraph Gateway integration
llm = ChatOpenAI(
    model="gpt-4o",
    temperature=0.2,
    base_url="https://gateway.aipromptshield.com/v1",
    default_headers={
        "X-APS-Policy": "strict-defense",
        "X-APS-PII-Masking": "presidio-strict",
        "X-APS-Agent-Sandbox": "ephemeral-microvm"
    }
)

# Multi-agent swarms execute with verified authority boundaries (OWASP ASI-01)
chain = prompt | llm
result = chain.invoke({"query": "Execute database summary"})

APS Gateway HTTP Policy Headers Reference

Header NameValuesDescription
X-APS-Policystrict-defense | audit-onlyControls whether attacks are blocked (403) or flagged for telemetry without dropping the query.
X-APS-PII-Maskingpresidio-strict | offEnables Microsoft Presidio NER entity scrubbing (SSN, credit cards, emails, secret tokens).
X-APS-Agent-Sandboxephemeral-microvm | standardIsolates autonomous agent tool calls (SQL, shell, webhooks) inside ephemeral Firecracker MicroVM containers.
X-APS-Evidence-Chaineu-article12-sha256Attaches cryptographic SHA-256 hash provenance receipts to HTTP response headers.