Hosted MCP Tools for AI Agent Orchestration

5 Security-Focused Model Context Protocol Servers

Deployed on Heroku • Integrated with Claude 3.7 • Ready for Production

This repository contains a presentation about deploying and integrating 5 hosted MCP (Model Context Protocol) tools with the Heroku Inference Addon for enhanced AI agent orchestration.

🚀 Live Presentation

View the presentation: https://benjaminrust.github.io/mcp-tools-presentation/

📋 Presentation Overview

What We Built

• GitHub Security MCP - Repository vulnerability analysis
• Brave Search Threat Intelligence - CVE database and threat search
• Python Execution Network Analysis - Security data analysis and risk scoring
• Slack Incident Response - Security alerts and team coordination
• ArXiv Academic Security Research - Security paper search and trend analysis

Architecture

All tools use FastMCP pattern with HTTP endpoints for seamless integration with the Heroku Inference Addon.

Technical Implementation

• FastMCP Pattern from the official MCP library
• Deployment: Heroku with Python 3.11.6 on Heroku-24 stack
• Dependencies: mcp==1.11.0, fastapi==0.116.1, uvicorn==0.35.0
• Integration: Claude 3.7 Inference Addon

🛠️ MCP Tools Deployed

1. GitHub Security MCP

• URL: github-mcp-tool-brust-547dc3ab814c.herokuapp.com
• Capabilities: Search security repositories, analyze repository security, trending security repos
• Use Cases: Vulnerability research, security code review, threat intelligence gathering

2. Brave Search Threat Intelligence MCP

• URL: brave-search-mcp-tool-brust-08cf6b0b7afb.herokuapp.com
• Capabilities: Search threat intelligence, CVE database queries, security news search
• Use Cases: CVE tracking and analysis, threat hunting, security awareness

3. Python Execution Network Analysis MCP

• URL: python-exec-mcp-tool-brust-c1ed7e307b16.herokuapp.com
• Capabilities: Analyze network data, calculate risk scores, generate security reports
• Use Cases: Network security monitoring, risk assessment, automated security reporting

4. Slack Incident Response MCP

• URL: slack-mcp-tool-brust-9dfb030fe646.herokuapp.com
• Capabilities: Send security alerts, coordinate incident response, get team status
• Use Cases: Incident response automation, security team coordination, real-time communication

5. ArXiv Academic Security Research MCP

• URL: arxiv-mcp-tool-brust-new-2d3e9a1d0e69.herokuapp.com
• Capabilities: Search security papers, analyze research trends, get paper summaries
• Use Cases: Academic research, security trend analysis, literature review automation

🔗 Integration with Heroku Inference Addon

Configuration Steps

1. Access Inference Addon Dashboard

   https://addons-sso.heroku.com/apps/network-intelligence-prod/addons/claude-3-7
   

2. Add MCP Server URLs

   • GitHub: https://github-mcp-tool-brust-547dc3ab814c.herokuapp.com/mcp
   • Brave Search: https://brave-search-mcp-tool-brust-08cf6b0b7afb.herokuapp.com/mcp
   • Python Execution: https://python-exec-mcp-tool-brust-c1ed7e307b16.herokuapp.com/mcp
   • Slack: https://slack-mcp-tool-brust-9dfb030fe646.herokuapp.com/mcp
   • ArXiv: https://arxiv-mcp-tool-brust-new-2d3e9a1d0e69.herokuapp.com/mcp

📊 Example Use Cases

Security Incident Response

1. Detect Threat → Brave Search MCP finds CVE information
2. Analyze Code → GitHub MCP searches for vulnerable patterns
3. Calculate Risk → Python Execution MCP assesses impact
4. Alert Team → Slack MCP coordinates response
5. Research Solutions → ArXiv MCP finds academic solutions

Threat Intelligence Gathering

1. Search Threats → Brave Search MCP queries threat databases
2. Analyze Patterns → Python Execution MCP processes data
3. Find Related Research → ArXiv MCP discovers academic context
4. Monitor Repositories → GitHub MCP tracks code changes
5. Update Team → Slack MCP shares findings

🏗️ Architecture Benefits

Scalability

• Independent Scaling - Each MCP tool scales separately
• Resource Isolation - No interference between tools
• Load Distribution - Requests distributed across apps

Maintainability

• Modular Design - Each tool is self-contained
• Easy Updates - Deploy individual tools independently
• Clear Separation - Network intelligence app remains untouched

Reliability

• Fault Isolation - Single tool failure doesn't affect others
• Redundancy - Multiple tools can provide similar functionality
• Monitoring - Individual tool health monitoring

📚 Resources

• MCP Specification
• Heroku Inference Addon
• FastMCP Documentation
• Slidev Documentation

👨‍💻 Author

Benjamin Rust - brust@salesforce.com

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Built with Slidev - Presentation slides for developers