Skip to main content
Glama

HexStrike AI MCP Server

by b-bogus
SETUP_DIRECT_MCP.md4.99 kB
# HexStrike AI Direct MCP Server Setup This guide explains how to set up the new direct network MCP server that eliminates the need for local client files. ## Architecture ``` Machine .16 (Claude Code) → Machine .71 (MCP Server + Flask API) ↓ MCP over network ↓ local HTTP calls Claude Code in VS Code hexstrike_mcp_server.py:8889 ↓ hexstrike_server.py:8888 ↓ Security Tools ``` ## Setup Instructions ### On Remote Machine (.71) - Server Side 1. **Install dependencies** (if not already done): ```bash # On Debian/Ubuntu sudo apt update sudo apt install python3-pip python3-requests python3-fastmcp # Or via pip pip3 install requests fastmcp flask psutil ``` 2. **Start the Flask API server** (existing): ```bash cd /path/to/hexstrike-ai python3 hexstrike_server.py ``` 3. **Start the new MCP server** (in another terminal): ```bash python3 hexstrike_mcp_server.py --host 0.0.0.0 --port 8889 ``` **Expected output:** ``` ✅ Connected to HexStrike API at http://localhost:8888 ✅ 12 tools available 🚀 Starting HexStrike Direct MCP Server 📡 Listening on 0.0.0.0:8889 🔗 API Backend: http://localhost:8888 ⚡ Configure Claude Code to connect to: http://0.0.0.0:8889 ``` ### On Local Machine (.16) - Client Side **Configure VS Code User Settings:** 1. Open VS Code 2. `Ctrl+Shift+P` → "Preferences: Open User Settings (JSON)" 3. Add this configuration: ```json { "git.enableSmartCommit": true, "mcp.servers": { "hexstrike-ai": { "command": "stdio", "args": [], "env": { "MCP_SERVER_URL": "http://192.168.0.71:8889" }, "description": "HexStrike AI Direct Network MCP Server", "timeout": 300 } } } ``` **Alternative Configuration (if the above doesn't work):** ```json { "git.enableSmartCommit": true, "mcp.servers": { "hexstrike-ai": { "command": "curl", "args": ["-X", "POST", "http://192.168.0.71:8889/mcp"], "description": "HexStrike AI Direct Network MCP Server", "timeout": 300 } } } ``` ## Testing the Connection ### From Machine .16 (Local): ```bash # Test basic connectivity curl http://192.168.0.71:8889/health # Test MCP endpoint (once server is running) curl -X POST http://192.168.0.71:8889/mcp \ -H "Content-Type: application/json" \ -d '{"method": "tools/list"}' ``` ### From Machine .71 (Remote): ```bash # Test Flask API is running curl http://localhost:8888/health # Test MCP server is running curl http://localhost:8889/health ``` ## Available Tools The MCP server exposes these tool categories: **Network & Reconnaissance:** - `nmap_scan()` - Advanced port scanning - `rustscan_scan()` - Fast port scanning - `amass_enum()` - Subdomain enumeration - `subfinder_scan()` - Passive subdomain discovery **Web Application Security:** - `gobuster_scan()` - Directory enumeration - `nuclei_scan()` - Vulnerability scanning - `sqlmap_scan()` - SQL injection testing - `httpx_scan()` - HTTP probing **Binary Analysis:** - `ghidra_analyze()` - Reverse engineering - `radare2_analyze()` - Binary analysis - `gdb_debug()` - Debugging - `volatility_analyze()` - Memory forensics **Cloud Security:** - `prowler_assess()` - Cloud security assessment - `trivy_scan()` - Container scanning - `kube_hunter_scan()` - Kubernetes testing **AI Intelligence:** - `ai_analyze_target()` - AI-powered target analysis - `ai_select_tools()` - Intelligent tool selection - `bugbounty_reconnaissance()` - Bug bounty workflows - `ctf_solve_challenge()` - CTF automation ## Usage Example Once configured, you can use Claude Code normally: ``` User: "I need to scan example.com for open ports" Claude: I'll use the nmap_scan tool to scan example.com for open ports. ``` Claude Code will automatically call the MCP server on .71, which will execute the scan and return results. ## Troubleshooting **Connection refused:** - Check if ports 8888 and 8889 are open on .71 - Verify firewall settings - Test with `telnet 192.168.0.71 8889` **MCP server won't start:** - Install missing dependencies: `pip3 install fastmcp requests` - Check if Flask server (port 8888) is running first **No tools available:** - Check if security tools are installed on .71 - Review `/health` endpoint for tool availability **Claude Code can't connect:** - Verify VS Code MCP configuration - Check Claude Code logs for connection errors - Try alternative configuration format ## Benefits ✅ **No local files needed** on machine .16 ✅ **True network MCP server** - proper remote access ✅ **Preserves existing architecture** - doesn't modify original files ✅ **Easy maintenance** - all tools and updates on one machine ✅ **Proper separation of concerns** - MCP server vs Flask API

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/b-bogus/hexstrike-ai_mcp_server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server