Chat Security Filter Extension for VS Code

🛡️ Automatically detect and filter sensitive information (credentials, infrastructure details, personal data) from AI chat messages before they're sent to language models. Protects your Azure, AWS, GCP resources, on-premise infrastructure, and secrets from accidental exposure.

🌟 Features

Real-time Scanning: Automatically scans all chat messages for sensitive information
Comprehensive Detection: Detects 80+ types of sensitive data including:

Cloud Credentials:
- AWS Access Keys, Secret Keys, Account IDs
- Azure Storage Keys, Client Secrets, Tenant IDs, Subscription IDs
- Google Cloud API Keys, OAuth tokens, Project IDs
- GitHub/GitLab Personal Access Tokens
- Generic API keys and Bearer tokens
Azure Infrastructure:
- App Service URLs (*.azurewebsites.net)
- App Service Environments (ASE)
- Service Bus namespaces (*.servicebus.windows.net)
- Storage Accounts (*.blob/table/queue/file.core.windows.net)
- Key Vault URLs (*.vault.azure.net)
- Azure SQL Servers (*.database.windows.net)
- Cosmos DB endpoints (*.documents.azure.com)
- Function Apps, Event Hubs, Redis Cache
- Container Registry (*.azurecr.io)
- Resource IDs, Subscription IDs, Tenant IDs
- Application Insights Instrumentation Keys
AWS Infrastructure (16 patterns):
- S3 Buckets, ARN (Amazon Resource Names)
- EC2 Instances, Lambda Functions, RDS Databases
- DynamoDB, ECS, EKS Clusters
- CloudFront, Elastic Beanstalk, API Gateway
- ElastiCache, SNS, SQS, Kinesis
- ECR Repositories, Account IDs
GCP Infrastructure (13 patterns):
- Project IDs, Service Accounts
- Cloud Storage, Cloud Run, Cloud Functions
- App Engine, Compute Engine, BigQuery
- Cloud SQL, Firestore/Firebase
- Pub/Sub, GKE Clusters, Artifact Registry
IBM Cloud Infrastructure (9 patterns):
- Cloud Resource Names (CRN)
- Cloud Foundry Apps, Object Storage
- Kubernetes Service, Cloud Functions
- Databases, Event Streams (Kafka)
- Watson Services, Account IDs
Alibaba Cloud Infrastructure (10 patterns):
- OSS Buckets, ECS Instances, RDS
- Server Load Balancer, Function Compute
- API Gateway, Container Registry
- ACK Clusters, Account IDs, RAM Roles
On-Premise Infrastructure:
- UNC paths (\\server\share)
- SQL Server connection strings
- Internal URLs with ports
- LDAP connection strings
- Windows domain names
- Private IP addresses (10.x, 172.x, 192.168.x)
Container & Kubernetes:
- Docker Registry URLs (GCR, ECR, ACR)
- Kubernetes cluster endpoints
Database Connections:
- MongoDB, PostgreSQL, MySQL connection strings with credentials
Security Artifacts:
- Private keys (RSA, SSH, OpenSSH)
- JWT tokens
- Slack tokens
Network Infrastructure:
- IPv4 addresses
- Internal domains (.local, .internal, .corp)
Personal Data:
- Email addresses and phone numbers
- Credit card numbers
Code Secrets:
- Passwords in code
- High-entropy strings (unknown tokens)
Flexible Actions:
- Auto-redact sensitive data with masking
- Block messages entirely (configurable)
- Preview filtered messages before sending
User Controls:
- /scan command: Scan text without AI processing
- /configure command: Quick access to settings
- Toggle filter on/off
- Adjustable detection sensitivity (low/medium/high)

🚀 Quick Start

Prerequisites

VS Code version 1.85.0 or higher
Node.js 18.x or higher
npm or yarn

Installation for Development/Testing

Clone or create the project:
```
cd c:\workspace\idea--vscode
```
Install dependencies:
```
npm install
```
Compile the TypeScript code:
```
npm run compile
```
Open in VS Code:
```
code .
```
Press F5 to launch Extension Development Host
- This opens a new VS Code window with your extension loaded
- The original window is for development
- The new window is for testing

🧪 Testing Your Extension

Method 1: Using Chat Participant (Recommended)

In the Extension Development Host window, open the Chat view (Ctrl+Alt+I or Cmd+Alt+I on Mac)
Type @secure to invoke your security filter participant

Try these example prompts:

@secure Check this code: My AWS key is AKIAIOSFODNN7EXAMPLE

@secure /scan postgresql://user:password123@localhost:5432/mydb

@secure /configure

Method 2: Testing Scan Command

Try scanning various text samples:

@secure /scan Here's my connection: mongodb://admin:secretPass@192.168.1.100:27017/production

Expected output: Detection of MongoDB credentials and IP address

Method 3: Testing Configuration

@secure /configure

This shows current settings and allows you to adjust sensitivity.

📋 Example Test Cases

Test Case 1: AWS Credentials

Input:

@secure I need help with this AWS config:
AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE
AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY

Expected: Both credentials detected and redacted

Test Case 2: GitHub Token

Input:

@secure Clone the repo using: git clone https://ghp_abcdef1234567890ABCDEFGHIJKLMNOPQR@github.com/myrepo.git

Expected: GitHub token detected and masked

Test Case 3: Database Connection

Input:

@secure Connect to: mysql://dbuser:superSecretPass@prod-db.company.internal:3306/app

Expected: MySQL credentials, password, and internal domain detected

Test Case 4: API Key in Code

Input:

@secure 
const config = {
  apiKey: "sk_test_EXAMPLE1234567890abcdefghijklmnop",
  apiSecret: "Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.EXAMPLE_PAYLOAD.EXAMPLE_SIGNATURE"
}

Expected: API key and JWT token detected

⚙️ Configuration

Access settings via:

Command: @secure /configure
Or: File → Preferences → Settings → Search "Chat Security Filter"

Available Settings

Setting	Type	Default	Description
`chatSecurityFilter.enabled`	boolean	`true`	Enable/disable automatic filtering
`chatSecurityFilter.blockOnDetection`	boolean	`false`	Block messages with credentials vs. auto-redact
`chatSecurityFilter.showNotifications`	boolean	`true`	Show warnings when sensitive data detected
`chatSecurityFilter.detectionLevel`	string	`medium`	Sensitivity: `low`, `medium`, or `high`

Example: Change Detection Level

Open Settings (Ctrl+,)
Search for "chatSecurityFilter.detectionLevel"
Select "high" for maximum security (more false positives)
Select "low" for fewer warnings (may miss some secrets)

🔧 Development Commands

# Compile TypeScript
npm run compile

# Watch mode (auto-compile on changes)
npm run watch

# Run tests
npm test

# Lint code
npm run lint

🐛 Debugging

Set breakpoints in src/extension.ts or src/securityFilter.ts
Press F5 to start debugging
Use Debug Console in the development window to see logs
Check Output → "Extension Host" for extension logs

Common Issues

Issue: Extension doesn't activate

Solution: Check package.json activation events, ensure VS Code version ≥ 1.85.0

Issue: Chat participant not showing

Solution: Verify contributes.chatParticipants in package.json, restart Extension Development Host

Issue: TypeScript errors

Solution: Run npm install again, check tsconfig.json

📦 Building for Distribution

Install vsce (VS Code Extension Manager):
```
npm install -g @vscode/vsce
```
Package the extension:
```
vsce package
```
This creates a .vsix file you can:
- Install locally: code --install-extension chat-security-filter-0.1.0.vsix
- Share with others
- Publish to the marketplace

🚀 Publishing to Marketplace

Create a publisher account: https://marketplace.visualstudio.com/manage
Get a Personal Access Token from Azure DevOps
Login with vsce:
```
vsce login your-publisher-name
```
Publish:
```
vsce publish
```

🔐 Security Considerations

Local Processing: All scanning happens locally, no data sent to external services
Pattern Updates: Regularly update detection patterns for new threat types
False Positives: High detection levels may flag legitimate data
Not 100% Foolproof: Should be part of a layered security approach

🛠️ Customization

Adding New Detection Patterns

Edit src/detectionPatterns.ts:

{
  name: 'My Custom Token',
  pattern: /my_token_[A-Za-z0-9]{32}/gi,
  description: 'Custom token format detected',
  severity: 'high',
  category: 'credential'
}

Changing Redaction Style

Edit src/securityFilter.ts, modify maskString() method for custom masking.

📊 Detection Statistics

The extension tracks:

Number of scans performed
Types of credentials detected
False positive feedback (planned)

View in Debug Console during development.

🤝 Contributing

Fork the repository
Create a feature branch
Add tests for new patterns
Submit a pull request

📄 License

MIT License - See LICENSE file for details

🔗 Resources

📞 Support

Report issues on GitHub
Questions: Use GitHub Discussions
Security concerns: Email security@yourcompany.com

Note: This is a development version. Test thoroughly before using in production environments.

Next Steps

✅ Install dependencies: npm install
✅ Compile: npm run compile
✅ Press F5 to test
✅ Try: @secure /scan <your sensitive text>
🚀 Enjoy secure AI chats!

Chat Security Filter

Rajadi Global Private Limited

Chat Security Filter Extension for VS Code

🌟 Features

🚀 Quick Start

Prerequisites

Installation for Development/Testing

🧪 Testing Your Extension

Method 1: Using Chat Participant (Recommended)

Method 2: Testing Scan Command

Method 3: Testing Configuration

📋 Example Test Cases

Test Case 1: AWS Credentials

Test Case 2: GitHub Token

Test Case 3: Database Connection

Test Case 4: API Key in Code

⚙️ Configuration

Available Settings

Example: Change Detection Level

🔧 Development Commands

🐛 Debugging

Common Issues

📦 Building for Distribution

🚀 Publishing to Marketplace

🔐 Security Considerations

🛠️ Customization

Adding New Detection Patterns

Changing Redaction Style

📊 Detection Statistics

🤝 Contributing

📄 License

🔗 Resources

📞 Support

Next Steps