Mirror Security Logo
Getting Started

Platform Overview

Architecture and components of the Mirror Security Platform

Mirror Security Platform Overview

The Mirror Security Platform provides a comprehensive security framework for GenAI applications, addressing the unique challenges of securing AI systems while maintaining their functionality and performance.

Platform Architecture

Our platform consists of four main products that can be used independently or together for complete protection:

  1. Code Prism: Advanced code analysis and security scanning platform for secure software development.
  2. Mirror Vectax SDK: Provides data-level security through encryption.
  3. AgentIQ Policy Engine: Delivers comprehensive AI safety, compliance, and application-level security through policy enforcement.
  4. Discover: GenAI Security Scanner and Automated RedTeaming Suite.

Code Prism

Code Prism provides advanced code analysis and security scanning to identify vulnerabilities and enforce coding standards.

Code Prism focuses on secure software development through comprehensive code analysis:

Vulnerability Detection

Identify security vulnerabilities, code smells, and potential security risks in your codebase.

Coding Standards

Enforce coding standards and best practices to ensure consistent, maintainable code.

Security Scanning

Comprehensive security analysis including SAST, dependency scanning, and configuration analysis.

Compliance Monitoring

Ensure compliance with security frameworks and regulatory requirements.

Key Technologies

  • Static Application Security Testing (SAST): Analyze source code for security vulnerabilities
  • Dependency Scanning: Identify vulnerable dependencies and outdated packages
  • Configuration Analysis: Review security configurations and settings
  • Compliance Mapping: Map findings to security frameworks and standards

Mirror Vectax SDK

Mirror Vectax SDK enables privacy-preserving AI by providing encryption designed for machine learning operations.

Vectax SDK focuses on data protection with cryptographic techniques designed specifically for AI operations:

Secure Vector Operations

Enterprise-grade security for vector operations with similarity-preserving encryption. Protect your embeddings while maintaining searchability.

Fine-Grained Access Control

Built-in RBAC with multi-dimensional policies. Control access at role, group, and department levels with comprehensive audit trails.

Format-Preserving Encryption

Encrypt sensitive data while maintaining format and searchability. Perfect for securing metadata, PII, and structured data.

Secure Agents Pipeline

End-to-end security for machine learning workflows. From data ingestion to model serving, protect every step of your ML pipeline.

Key Technologies

  • Similarity-Preserving Encryption (SPE): Allows encrypted vector similarity search
  • Format-Preserving Encryption (FPE): Maintains data format while encrypting
  • Attribute-Based Access Control (ABAC): Enables fine-grained permissions
  • Secure Multi-party Computation: Enables collaborative AI without sharing raw data

AgentIQ Policy Engine

The AgentIQ Policy Engine provides comprehensive AI safety, governance, security, and compliance controls for AI applications.

AgentIQ Policy Engine focuses on AI safety and application-level security through advanced detection, monitoring, and policy enforcement:

AI Safety & Threat Detection

Detect and prevent security threats like prompt injection, jailbreaks, and adversarial attacks in real-time.

Content Moderation

Analyze content for toxicity, bias, and inappropriate material with high accuracy and context awareness.

Privacy Protection

Automatically identify and protect PII, ensuring GDPR/HIPAA compliance and data privacy.

Quality Assurance

Evaluate RAG quality, detect hallucinations, and ensure response accuracy with ground truth validation.

Declarative Policy Grammar

Define policies using a simple, readable grammar that's both powerful and intuitive.

Real-time Enforcement

Apply policies at runtime to prevent violations before they occur.

Comprehensive Protection

Built-in rules for PII protection, prompt safety, RAG security, and more.

Seamless Integration

Apply policies through simple decorators with minimal code changes.

Key Technologies

  • AI Safety Technologies: Prompt injection detection, content analysis, PII detection, hallucination detection, and RAG quality assessment
  • Policy Definition Language: Simple yet powerful grammar for defining security rules
  • Runtime Inspection: Real-time inspection of AI inputs and outputs
  • Compliance Mapping: Automatic mapping of policies to regulatory frameworks
  • Telemetry Collection: Comprehensive logging and auditing capabilities

Integration Points

The Mirror Security Platform integrates with your AI ecosystem at multiple levels:

Code Security Layer: Code Prism analyzes and secures your codebase, identifying vulnerabilities and enforcing coding standards.

Data Layer: Vectax SDK secures data at rest and in transit, protecting embeddings, metadata, and model weights.
Application Layer: AgentIQ Policy Engine provides comprehensive AI safety, compliance, and policy enforcement, monitoring inputs and outputs to AI models.

Infrastructure Layer: Platform-wide capabilities ensure consistent security across your entire AI infrastructure.

Supported Integrations

  • ChromaDB
  • Pinecone
  • Weaviate
  • Milvus
  • Qdrant
  • Mongo
  • Vespa

Deployment Options

Mirror Security Platform offers flexible deployment options to meet your specific needs:

Cloud SaaS

Fully managed service with no infrastructure maintenance

Private Cloud

Dedicated deployment in your private cloud environment

On-Premises

Self-hosted deployment within your own data center

Next Steps

Code Prism

Discover code analysis and security scanning

Quick Start Guide

Get up and running with Mirror Security in minutes

Vectax SDK

Explore the security features of Vectax SDK

AgentIQ Policy Engine

Learn how to define and enforce AI security policies

Sign Up for Free

See Mirror Security in action with a personalized demo

Previous

Introduction

Next

Quick Start Guide

On this page