🚨 Active Supply Chain Attack:node-ipc Package Compromised.Learn More
Socket
Book a DemoSign in
Socket

@aegis-sdk/openai

Package Overview
Dependencies
Maintainers
1
Versions
3
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

@aegis-sdk/openai

OpenAI SDK adapter for Aegis prompt injection defense

latest
npmnpm
Version
0.5.0
Version published
Weekly downloads
12
-36.84%
Maintainers
1
Weekly downloads
 
Created
Source

@aegis-sdk/openai

OpenAI SDK adapter for Aegis prompt injection defense. Scans messages, monitors streams, and validates tool calls in the OpenAI chat completion format.

Part of the Aegis.js prompt injection defense toolkit.

Installation

npm install @aegis-sdk/openai @aegis-sdk/core openai

Quick Start

Wrap the OpenAI client for automatic protection on every call:

import OpenAI from 'openai';
import { Aegis } from '@aegis-sdk/core';
import { wrapOpenAIClient } from '@aegis-sdk/openai';

const aegis = new Aegis({ policy: 'strict' });
const client = wrapOpenAIClient(new OpenAI(), aegis);

// Messages are scanned before sending.
// Streaming responses are monitored in real-time.
// Tool calls are validated against the policy.
const stream = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello!' }],
  stream: true,
});

Or scan messages manually:

import { Aegis } from '@aegis-sdk/core';
import { guardMessages } from '@aegis-sdk/openai';

const aegis = new Aegis({ policy: 'strict' });

const messages = [
  { role: 'system' as const, content: 'You are a helpful assistant.' },
  { role: 'user' as const, content: userInput },
];

// Throws AegisInputBlocked if injection is detected
const safe = await guardMessages(aegis, messages);

API

wrapOpenAIClient(client, aegis, options?)

Proxy the OpenAI client to automatically guard all chat.completions.create() calls. Input messages are scanned before sending, streaming responses are monitored, and tool/function calls are validated against the Aegis policy. All other client methods pass through unchanged.

guardMessages(aegis, messages, options?)

Scan an array of OpenAIChatCompletionMessageParam[] for prompt injection. Extracts text from all message formats (string content, multi-modal content parts, tool/function messages) and runs them through aegis.guardInput(). Optionally validates tool_calls and function_call blocks against the policy (enabled by default). Returns the original messages if safe, throws AegisInputBlocked if blocked.

createStreamTransform(aegis)

Create a TransformStream<string, string> for monitoring extracted text deltas. Feed delta.content values from OpenAI streaming chunks through this transform for real-time output scanning.

getAuditLog(aegis)

Convenience accessor for the Aegis audit log.

Options

guardMessages and the client wrapper accept OpenAIGuardOptions:

  • scanStrategy -- 'last-user' (default), 'all-user', or 'full-history'
  • validateToolCalls -- Whether to validate tool/function calls against the policy (default: true)

Learn More

License

MIT

FAQs

Package last updated on 24 Feb 2026

Did you know?

Socket

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts