@evalguardai/openai

@evalguard/openai

Drop-in OpenAI SDK wrapper that adds real-time guardrails, trace logging, and cost tracking via EvalGuard.

Installation

npm install @evalguard/openai openai

Quick Start

import OpenAI from "openai";
import { wrapOpenAI } from "@evalguard/openai";

const openai = wrapOpenAI(new OpenAI(), {
  apiKey: "eg_...",
  projectId: "proj_...",
});

// Use exactly like the normal OpenAI SDK — guardrails are automatic
const response = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello, how are you?" }],
});

console.log(response.choices[0].message.content);

Streaming

Streaming works transparently. The wrapper intercepts chunks to log the assembled response without affecting stream behavior.

const stream = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Write a poem about AI safety" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

Configuration

const openai = wrapOpenAI(new OpenAI(), {
  // Required: your EvalGuard API key
  apiKey: "eg_...",

  // Optional: EvalGuard API base URL (default: https://evalguard.ai/api/v1)
  baseUrl: "https://your-evalguard-instance.com/api/v1",

  // Optional: block requests that fail guardrails (default: true)
  blockOnViolation: true,

  // Optional: log all requests to EvalGuard (default: true)
  enableLogging: true,

  // Optional: project ID for organizing traces
  projectId: "proj_...",

  // Optional: custom metadata attached to every trace
  metadata: { environment: "production", service: "chatbot" },

  // Optional: callback when a guardrail violation is detected
  onViolation: (result) => {
    console.warn("Guardrail violation:", result.violations);
  },
});

What It Does

Phase	Action
Pre-request	Sends the prompt to EvalGuard's firewall for prompt injection detection, PII scanning, and toxicity checks
LLM call	Passes through to the real OpenAI API unchanged
Post-response	Logs model, tokens, latency, cost, and guardrail results as a trace to EvalGuard

Fail-Open Design

If EvalGuard is unreachable (network error, timeout, 5xx), the wrapper passes requests through to OpenAI directly. Your application never breaks because of EvalGuard downtime.

Error Handling

When blockOnViolation is true (default) and a guardrail check fails:

import { EvalGuardViolationError } from "@evalguard/openai";

try {
  const response = await openai.chat.completions.create({
    model: "gpt-4o",
    messages: [{ role: "user", content: "malicious prompt..." }],
  });
} catch (error) {
  if (error instanceof EvalGuardViolationError) {
    console.log("Blocked:", error.violations);
    // [{ type: "prompt_injection", severity: "critical", message: "..." }]
  }
}

Set blockOnViolation: false to log violations without blocking:

const openai = wrapOpenAI(new OpenAI(), {
  apiKey: "eg_...",
  blockOnViolation: false,
  onViolation: (result) => {
    // Log but don't block
    analytics.track("guardrail_violation", result);
  },
});

License

MIT