Security Architecture

Comprehensive security architecture for Cursor IDE with Azure AI Foundry integration.

Table of contents

1. Overview
   1. Security Principles
2. Architecture Layers
3. Key Components
   1. 1. Identity & Access Management
   2. 2. Network Security
   3. 3. Data Protection
   4. 4. Application Security
   5. 5. Monitoring & Response
4. Security Boundaries
   1. Trust Boundaries
5. Data Flow Diagram
   1. Data Flow Security Controls
6. Threat Model
   1. Identified Threats & Mitigations
   2. Attack Scenarios & Defenses
7. Compliance Mapping
   1. Framework Alignment
   2. Compliance Evidence Collection
8. Security Validation
   1. Automated Security Checks
9. Next Steps

---

Overview

This section documents the end-to-end security architecture for deploying Cursor IDE in enterprise environments with Azure AI Foundry integration. The architecture implements defense-in-depth principles and zero-trust security models.

Security Principles

Principle	Implementation
Zero Trust	Never trust, always verify - authenticate every request
Least Privilege	RBAC ensures minimum necessary access
Defense in Depth	Multiple security layers protect data
Assume Breach	Network segmentation limits blast radius
Data Sovereignty	All processing stays within Azure tenant

---

Architecture Layers

graph TB
    subgraph "Identity Layer"
        A[Azure Entra ID] --> B[Conditional Access]
        B --> C[MFA]
    end
    
    subgraph "Network Layer"
        D[Private Endpoints] --> E[Azure VNet]
        E --> F[NSGs]
    end
    
    subgraph "Application Layer"
        G[Cursor IDE] --> H[Privacy Mode]
        H --> I[API Client]
    end
    
    subgraph "Data Layer"
        J[Azure OpenAI] --> K[Encryption at Rest]
        K --> L[TLS 1.3 in Transit]
    end
    
    subgraph "Monitoring Layer"
        M[Azure Monitor] --> N[Log Analytics]
        N --> O[Azure Sentinel]
    end
    
    A --> G
    D --> J
    G --> D
    J --> M
    
    style A fill:#00a6ed,stroke:#0086bd,color:#fff
    style G fill:#7c3aed,stroke:#5b21b6,color:#fff
    style J fill:#0066cc,stroke:#003d7a,color:#fff
    style M fill:#10b981,stroke:#059669,color:#fff

---

Key Components

1. Identity & Access Management

• Azure Entra ID: Centralized identity provider
• Conditional Access: Policy-based access control
• RBAC: Role-based permissions on Azure resources
• MFA: Multi-factor authentication required

2. Network Security

• Private Endpoints: No public internet exposure
• VNet Integration: Traffic stays on Azure backbone
• NSGs: Network-level filtering
• Azure Firewall: Egress control

3. Data Protection

• Encryption at Rest: Azure Storage Service Encryption
• Encryption in Transit: TLS 1.3 mandatory
• Key Management: Azure Key Vault
• Data Residency: Regional deployment control

4. Application Security

• Privacy Mode: Zero data retention in Cursor
• Secret Management: No secrets in code or AI chat
• Extension Control: Allowlist-based extensions
• Audit Logging: Comprehensive activity tracking

5. Monitoring & Response

• Azure Monitor: Real-time metrics and logs
• Log Analytics: Centralized log storage (2+ years)
• Azure Sentinel: SIEM and threat detection
• Automated Alerts: Immediate incident notification

---

Security Boundaries

Trust Boundaries

┌─────────────────────────────────────────────────────────┐
│ Untrusted Zone: Public Internet                         │
│  - External attackers                                   │
│  - No direct access to Azure resources                  │
└─────────────────────────────────────────────────────────┘
                          ↕️ (Firewall)
┌─────────────────────────────────────────────────────────┐
│ Trusted Zone: Azure Tenant                               │
│  ┌──────────────────────────────────────────────────┐   │
│  │ Developer Workstations (Managed, MFA-enforced)   │   │
│  │  - Cursor IDE with Privacy Mode                  │   │
│  │  - Azure CLI authenticated                        │   │
│  └──────────────────────────────────────────────────┘   │
│                        ↕️ (Private Endpoints)            │
│  ┌──────────────────────────────────────────────────┐   │
│  │ Azure OpenAI Service (Private Network Only)      │   │
│  │  - Model inference                                │   │
│  │  - No data retention                              │   │
│  └──────────────────────────────────────────────────┘   │
│                        ↕️ (RBAC)                         │
│  ┌──────────────────────────────────────────────────┐   │
│  │ Azure Key Vault (Secrets Storage)                │   │
│  │  - API keys                                       │   │
│  │  - Certificates                                   │   │
│  └──────────────────────────────────────────────────┘   │
│                        ↕️ (Audit Stream)                 │
│  ┌──────────────────────────────────────────────────┐   │
│  │ Azure Monitor + Sentinel (SIEM)                  │   │
│  │  - 730-day retention                             │   │
│  │  - Real-time threat detection                    │   │
│  └──────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────┘

---

Data Flow Diagram

Detailed data flow showing every step of an AI completion request:

sequenceDiagram
    participant Dev as Developer
    participant Cursor as Cursor IDE
    participant EntraID as Azure Entra ID
    participant KV as Key Vault
    participant PE as Private Endpoint
    participant AOAI as Azure OpenAI
    participant Mon as Azure Monitor
    
    Dev->>Cursor: Type code, request completion
    Cursor->>EntraID: Authenticate user (SSO)
    EntraID-->>Cursor: JWT token
    
    Cursor->>KV: Request API key
    KV->>EntraID: Verify token
    EntraID-->>KV: Token valid
    KV-->>Cursor: API key (ephemeral)
    
    Cursor->>PE: HTTPS request (with API key)
    PE->>AOAI: Forward to Azure OpenAI (internal)
    AOAI->>AOAI: Process request (no storage)
    AOAI-->>PE: Completion response
    PE-->>Cursor: Return completion
    
    AOAI->>Mon: Log request metadata
    KV->>Mon: Log key access
    Cursor->>Mon: Log session activity
    
    Mon->>Mon: Analyze for anomalies
    
    Note over Dev,Mon: All data stays within Azure tenant

Data Flow Security Controls

Step	Security Control	Purpose
1. Developer Authentication	Azure Entra ID MFA	Verify user identity
2. Cursor to Entra ID	TLS 1.3	Encrypt authentication
3. Key Vault Access	RBAC + Conditional Access	Authorize secret access
4. API Request	Private Endpoint	Keep traffic on Azure network
5. Azure OpenAI Processing	No data retention	Prevent AI training on enterprise code
6. Response	TLS 1.3	Encrypt completion
7. Audit Logging	Azure Monitor	Create immutable audit trail

---

Threat Model

Identified Threats & Mitigations

Threat	Risk Level	Mitigation
API Key Exposure	🔴 Critical	Store in Key Vault, rotate monthly, audit access
Code Leakage to AI Provider	🔴 Critical	Privacy Mode ON, Azure tenant only, no external AI
Unauthorized Access	🟠 High	Entra ID + MFA + Conditional Access + RBAC
Network Eavesdropping	🟠 High	TLS 1.3, private endpoints, no public internet
Insider Threat	🟡 Medium	Comprehensive audit logs, least privilege RBAC
Supply Chain Attack	🟡 Medium	Extension allowlist, dependency scanning
Cost Overrun	🟢 Low	Budget alerts, rate limiting, quota management

Attack Scenarios & Defenses

Scenario 1: Attacker Steals API Key

Defense:

1. Key Vault access requires Entra ID auth (attacker can’t access)
2. RBAC limits who can read secrets
3. Key rotation means stolen key expires quickly
4. Azure Monitor alerts on unauthorized Key Vault access
5. Conditional Access can block suspicious IPs

Scenario 2: Malicious Extension Exfiltrates Code

Defense:

1. Extension allowlist prevents unapproved extensions
2. Network egress monitoring detects unusual traffic
3. Cursor Privacy Mode blocks telemetry
4. File exclusions (.cursorignore) protect sensitive files

Scenario 3: Compromised Developer Account

Defense:

1. MFA prevents credential-only compromise
2. Conditional Access blocks risky sign-ins
3. Audit logs show unusual activity
4. Just-in-time access limits blast radius
5. Azure Sentinel detects anomalous behavior

---

Compliance Mapping

Framework Alignment

Control	CIS Azure	NIST 800-53	ISO 27001	CSA CCM
Identity & Access	1.1-1.23	AC-2, IA-5	A.9.1, A.9.2	IAM-01
Network Security	6.1-6.6	SC-7	A.13.1	IVS-01
Data Protection	3.1-3.10	SC-28	A.10.1	EKM-02
Audit Logging	5.1-5.3	AU-2, AU-12	A.12.4	LOG-01
Incident Response	N/A	IR-4, IR-6	A.16.1	SEF-02

Compliance Evidence Collection

Evidence Type	Source	Retention
Access Logs	Azure Entra ID Sign-ins	2 years
API Logs	Azure OpenAI Diagnostics	2 years
Secret Access	Key Vault Audit Logs	7 years
Network Traffic	NSG Flow Logs	90 days
Security Alerts	Azure Monitor/Sentinel	2 years

---

Security Validation

Automated Security Checks

#!/bin/bash
# security-validation.sh

echo "=== Cursor + Azure AI Security Validation ==="

# Check 1: Privacy Mode enabled
echo "1. Checking Cursor Privacy Mode..."
if grep -q '"cursor.privacyMode": true' ~/.cursor/settings.json; then
    echo "   ✅ Privacy Mode enabled"
else
    echo "   ❌ Privacy Mode NOT enabled"
fi

# Check 2: No hardcoded secrets
echo "2. Scanning for hardcoded secrets..."
if ! grep -r "api[_-]key.*=" ~/.cursor/ 2>/dev/null; then
    echo "   ✅ No hardcoded API keys found"
else
    echo "   ❌ Potential API keys found"
fi

# Check 3: Private endpoint connectivity
echo "3. Verifying private endpoint..."
AOAI_IP=$(nslookup aoai-cursor-prod.openai.azure.com | grep Address | tail -1 | awk '{print $2}')
if [[ $AOAI_IP == 10.* ]]; then
    echo "   ✅ Private endpoint resolving correctly"
else
    echo "   ⚠️  Using public endpoint: $AOAI_IP"
fi

# Check 4: Audit logging enabled
echo "4. Checking Azure Monitor configuration..."
az monitor diagnostic-settings list \
    --resource /subscriptions/{sub-id}/resourceGroups/rg-cursor-ai-research/providers/Microsoft.CognitiveServices/accounts/aoai-cursor-prod \
    --query 'value[].logs[?enabled].category' -o tsv \
    | grep -q "Audit" && echo "   ✅ Audit logging enabled" || echo "   ❌ Audit logging NOT enabled"

echo "=== Validation Complete ==="

---

Next Steps

Explore detailed security documentation:

• Data Flow Diagrams - Visual architecture diagrams
• Tenant Isolation - Ensuring data sovereignty
• Compliance Considerations - Framework alignment details
• Threat Model - Complete threat analysis

---

Last Updated: October 10, 2025
Status: Security Reviewed

---

Table of contents