We release patches for security vulnerabilities for the following versions:

We take the security of AgentHub seriously. If you discover a security vulnerability, please follow these guidelines:

DO NOT open a public GitHub issue for security vulnerabilities.

Instead, please report security issues by:

1. Email: Send details to feedback@propstreet.com
2. GitHub Security Advisories: Use the "Security" tab on our GitHub repository to report privately

Please include as much of the following information as possible:

• Type of vulnerability (e.g., code execution, privilege escalation, data exposure)
• Affected components (e.g., HTTP transport, persistence, expert bridge)
• Reproduction steps - Clear, step-by-step instructions to reproduce the issue
• Impact assessment - How the vulnerability could be exploited
• Suggested fix - If you have one
• Your contact information - For follow-up questions

• Acknowledgment: Within 48 hours of report
• Initial assessment: Within 7 days
• Status updates: Every 7 days until resolved
• Fix timeline: Security patches prioritized based on severity

We use the following severity classifications:

Critical - Immediate action required

• Remote code execution
• Authentication bypass
• Data exposure of sensitive information

High - Fix within 7 days

• Privilege escalation
• Denial of service
• Information disclosure

Medium - Fix within 30 days

• Cross-site scripting (if applicable)
• Logic errors with security impact

Low - Fix in next release

• Minor information leaks
• Non-security-impacting bugs

1. Local-Only by Default

   • AgentHub is designed to run on localhost only
   • Never expose the HTTP endpoint to public networks
   • Use firewalls to restrict access if needed

2. Environment Variables

   • Never commit .env files to version control
   • Rotate Azure OpenAI API keys regularly
   • Use environment-specific credentials

3. Network Security

   • CORS is restricted to localhost origins only
   • No authentication required for local-only usage
   • If exposing to network, add authentication layer

1. Azure OpenAI Integration

   • Store API keys in environment variables, not in code
   • Use Azure Key Vault for production deployments
   • Enable API key rotation
   • Monitor usage for anomalies

2. Filesystem Watcher

   • Only watch directories you control
   • Be cautious with WATCH_ROOT in shared environments
   • Review ignore patterns for sensitive files

3. Persistence

   • .agenthub/ directory contains state snapshots
   • Ensure proper file permissions (user-only read/write)
   • Consider encrypting snapshots if storing sensitive data
   • Regularly clean up old snapshots

1. Message Content

   • Don't send sensitive credentials via agent messages
   • Review message history before sharing logs
   • Message bus is in-memory only (not persisted by default)

2. Expert Escalation

   • Code sent to Azure OpenAI may be stored by Microsoft
   • Don't escalate files containing secrets
   • Review Azure OpenAI data retention policies

3. Logs

   • LOG_LEVEL=debug may expose sensitive information
   • Sanitize logs before sharing
   • Rotate logs in production environments

1. No Authentication - AgentHub assumes trusted local environment
2. In-Memory State - No built-in encryption (use OS-level encryption)
3. HTTP Transport - No TLS required for localhost (use reverse proxy if needed)

• Local-Only: Default configuration prevents external access
• CORS Restrictions: Only localhost origins allowed
• Input Validation: All operations validated with Zod schemas
• Resource Limits: Configurable limits prevent resource exhaustion

We follow a coordinated disclosure process:

1. Private reporting to security team
2. Acknowledgment and investigation (48 hours)
3. Fix development (timeline based on severity)
4. Security advisory published after fix is available
5. Public disclosure after users have time to update (typically 7-14 days)

We will credit security researchers in:

• Security advisories
• Release notes
• CHANGELOG.md

Please let us know how you'd like to be credited (name, handle, company, or anonymous).

Subscribe to security updates:

• GitHub Watch: Enable "Security alerts" on the repository
• Release Notes: Check CHANGELOG.md for security patches
• GitHub Security Advisories: Follow our security advisories

• AgentHub server codebase (src/server/)
• Dashboard codebase (src/dashboard/)
• HTTP transport layer
• MCP protocol implementation
• Configuration and environment handling

• Third-party dependencies (report to upstream projects)
• Azure OpenAI service (report to Microsoft)
• MCP client implementations (report to client maintainers)
• Local development environment issues

For security concerns, contact:

• Email: feedback@propstreet.com
• GitHub: Use Security tab for private vulnerability reports

For general inquiries:

• Issues: https://github.com/propstreet/agenthub/issues
• Discussions: https://github.com/propstreet/agenthub/discussions

Last Updated: November 13, 2025

Thank you for helping keep AgentHub secure!