Skip to main content

Google Gemini Integration

KoreShield proxies Gemini requests through the OpenAI-compatible endpoint and applies sanitization and policy enforcement before forwarding to Gemini.

Installation​

npm install koreshield
pip install koreshield

Basic Request (TypeScript)​

import { Koreshield } from "koreshield";

const koreshield = new Koreshield({
  apiKey: process.env.KORESHIELD_API_KEY
});

const scan = await koreshield.scanPrompt("Summarize the quarterly update.");

if (!scan.isSafe) {
  throw new Error("Threat detected");
}

const response = await fetch("http://localhost:8000/v1/chat/completions", {
  method: "POST",
  headers: { "content-type": "application/json" },
  body: JSON.stringify({
    model: "gemini-1.5-pro",
    messages: [{ role: "user", content: "Summarize the quarterly update." }]
  })
});

Basic Request (Python)​

import os
import requests
from koreshield import KoreShieldClient

koreshield = KoreShieldClient(api_key=os.environ["KORESHIELD_API_KEY"])

scan = koreshield.scan_prompt("Summarize the quarterly update.")
if not scan.is_safe:
    raise Exception("Threat detected")

response = requests.post(
    "http://localhost:8000/v1/chat/completions",
    json={
        "model": "gemini-1.5-pro",
        "messages": [{"role": "user", "content": "Summarize the quarterly update."}]
    }
)

Streaming​

const response = await fetch("http://localhost:8000/v1/chat/completions", {
  method: "POST",
  headers: { "content-type": "application/json" },
  body: JSON.stringify({
    model: "gemini-1.5-pro",
    stream: true,
    messages: [{ role: "user", content: "Write a release summary." }]
  })
});
response = requests.post(
    "http://localhost:8000/v1/chat/completions",
    json={
        "model": "gemini-1.5-pro",
        "stream": True,
        "messages": [{"role": "user", "content": "Write a release summary."}]
    },
    stream=True
)

for line in response.iter_lines():
    if line:
        print(line.decode("utf-8"))

System Prompts and Multi-Turn​

{
  "model": "gemini-1.5-pro",
  "messages": [
    {"role": "system", "content": "You are a product analyst."},
    {"role": "user", "content": "Summarize the update."},
    {"role": "assistant", "content": "Summary..."},
    {"role": "user", "content": "List risks."}
  ]
}
payload = {
    "model": "gemini-1.5-pro",
    "messages": [
        {"role": "system", "content": "You are a product analyst."},
        {"role": "user", "content": "Summarize the update."},
        {"role": "assistant", "content": "Summary..."},
        {"role": "user", "content": "List risks."}
    ]
}

Compatibility Notes​

KoreShield forwards OpenAI-compatible payloads to the Gemini provider adapter. Validate any advanced fields with your Gemini setup before using in production.

Error Handling​

  • 403 indicates a blocked request due to policy enforcement.
  • 429 or 5xx typically indicates provider or rate-limit issues.

Security Controls​

security:
  sensitivity: medium
  default_action: block
  features:
    sanitization: true
    detection: true
    policy_enforcement: true

Next Steps​