How do I use AccuKnox MCP Server?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@AccuKnox MCP Server how many cloud assets do I have?" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

AccuKnox MCP Server - Testing Guide

Complete guide for testing MCP servers with stdio client, Gemini CLI, HTTP client, and VSCode Copilot.

📋 Table of Contents

Prerequisites

Python 3.10 or higher
pip package manager
Gemini CLI (optional, for AI integration)
VSCode or Cursor (optional, for IDE integration)

Project Structure

MCP_server/
├── MCP_server.py              # stdio server (for Gemini CLI)
├── MCP_server_http.py         # HTTP server (for web/remote)
├── clients/
│   ├── __init__.py
│   ├── stdio_client.py        # Test client for stdio
│   └── http_client.py         # Test client for HTTP
├── shared/
│   ├── __init__.py
│   ├── tools.py               # Tool implementations
│   └── api.py          # AccuKnox API client
├── .vscode/
│   └── mcp.json               # VSCode Copilot configuration
├── .env                       # Environment variables (create your own)
├── requirements.txt
└── README.md

Installation

1. Clone/Setup Project

cd MCP_server

2. Create Virtual Environment

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Configure Environment

Create .env file:

ACCUKNOX_BASE_URL=https://cspm.demo.accuknox.com
ACCUKNOX_API_TOKEN=your_token_here

Testing stdio Server

The stdio server uses standard input/output for communication, designed for local CLI integration.

With Test Client

Terminal 1: Start server (automatically started by client)

Terminal 2: Run test client

python3 clients/stdio_client.py

Interactive Menu:

Options:
  1 - Count assets
  2 - List 5 assets
  3 - Search by category
  4 - Get vulnerabilities
  exit - Quit

Select: 1

Example Output:

Total assets: 1247

With Gemini CLI

Gemini CLI provides AI-powered natural language interactions with your MCP server.

1. Install Gemini CLI

pip install google-generativeai

2. Configure MCP Server

# Add server to Gemini CLI (use absolute path)
gemini mcp add accuknox python3 /absolute/path/to/MCP_server.py

3. Verify Configuration

gemini mcp list

# Output:
# accuknox: python3 /path/to/MCP_server.py

4. Start Gemini CLI

gemini

5. Test with Natural Language Queries

Example 1: Count Assets

You: How many cloud assets do I have?

Gemini: You have 1,247 cloud assets in your inventory.

Example 2: Search by Category

You: Show me 5 Container assets

Gemini: Here are 5 Container assets:
1. nginx-prod (Container) - Region: us-east-1
2. redis-cache (Container) - Region: us-west-2
3. postgres-db (Container) - Region: eu-west-1
4. kafka-broker (Container) - Region: us-east-1
5. redis-sentinel (Container) - Region: ap-south-1

Example 3: Security Vulnerabilities

You: What security vulnerabilities do my AI models have?

Gemini: Your AI models have 147 security issues:
• ML Models: 5 issues (1 Critical, 4 Medium)
• LLM Models: 136 issues (136 Critical)
• Datasets: 6 issues (6 High)

Immediate action required for 137 critical vulnerabilities.

6. Remove Server (Optional)

gemini mcp remove accuknox

With VSCode Copilot (stdio)

1. Create Configuration File

Create .vscode/mcp.json in your project root:

mkdir -p .vscode
nano .vscode/mcp.json

2. Add stdio Server Configuration

{
  "mcpServers": {
    "accuknox-stdio": {
      "command": "python3",
      "args": ["${workspaceFolder}/MCP_server.py"],
      "env": {
        "ACCUKNOX_BASE_URL": "https://cspm.demo.accuknox.com",
        "ACCUKNOX_API_TOKEN": "your_token_here"
      }
    }
  }
}

Note: Using ${workspaceFolder} makes the configuration portable across machines.

3. Activate in VSCode

Open your project in VSCode
VSCode automatically detects .vscode/mcp.json
Click "Start" when prompted
Open Copilot Chat: Ctrl+Shift+I (Windows/Linux) or Cmd+Shift+I (Mac)
Enable Agent Mode (click brain icon )
Click tools icon () → Enable accuknox-stdio

4. Test with Copilot

@agent How many cloud assets do I have?
@agent Show me Container assets in AWS
@agent What are my model vulnerabilities?

Testing HTTP Server

The HTTP server provides REST API access for web clients and remote integrations.

With HTTP Test Client

Terminal 1: Start HTTP server

python3 MCP_server_http.py

Output:

======================================================================
AccuKnox MCP Server - HTTP
======================================================================
Server: http://localhost:8000
Tools: search_assets, get_model_vulnerabilities
Press Ctrl+C to shutdown
======================================================================
INFO:     Uvicorn running on http://0.0.0.0:8000

Terminal 2: Run test client

python3 clients/http_client.py

Interactive Menu:

Options:
  1 - Count assets
  2 - List 5 assets
  3 - Search by category
  4 - Get vulnerabilities
  exit - Quit

Select: 4

With VSCode Copilot (HTTP)

VSCode Copilot supports HTTP MCP servers starting from version 1.102+.

Prerequisites

VSCode 1.102 or higher
GitHub Copilot extension installed
HTTP MCP server running on http://localhost:8000

Step 1: Verify HTTP Server

# Start HTTP server
python3 MCP_server_http.py

# In another terminal, verify it's running
curl http://localhost:8000/health
# Expected: {"status": "healthy"}

Step 2: Configure VSCode

Option A: Project-Level Configuration (Recommended)

Create .vscode/mcp.json:

{
  "servers": {
    "accuknox-http": {
      "type": "http",
      "url": "http://localhost:8000"
    }
  },
  "inputs": []
}

Option B: Remote Server Configuration

For deployed servers:

{
  "servers": {
    "accuknox-http": {
      "type": "http",
      "url": "https://your-server.com:8000",
      "headers": {
        "Authorization": "Bearer ${input:api-token}"
      }
    }
  },
  "inputs": [
    {
      "type": "promptString",
      "id": "api-token",
      "description": "AccuKnox API Token",
      "password": true
    }
  ]
}

Option C: Global Configuration

For all workspaces, create ~/.vscode/mcp.json:

{
  "servers": {
    "accuknox-http": {
      "type": "http",
      "url": "http://localhost:8000"
    }
  },
  "inputs": []
}

Step 3: Activate Server

Method 1: Automatic Detection

Open project in VSCode
VSCode detects .vscode/mcp.json
Click "Start" button in the file
Server status shows "running" in status bar

Method 2: Command Palette

Press Ctrl+Shift+P (or Cmd+Shift+P on Mac)
Type: "MCP: Add Server"
Select "HTTP/SSE"
Enter Server ID: accuknox-http
Enter URL: http://localhost:8000
Choose Workspace or Global
Click Save

Method 3: Manual Configuration

Create .vscode/mcp.json with above content
Save file
VSCode auto-detects and shows Start button

Step 4: Enable Copilot Agent Mode

Open Copilot Chat:
- Press Ctrl+Shift+I (Windows/Linux)
- Press Cmd+Shift+I (Mac)
- Or click Copilot icon in Activity Bar
Enable Agent Mode:
- Click the brain icon () at top
- Or type @agent in chat
Configure Tools:
- Click tools icon () at bottom
- Find accuknox-http in list
- Toggle ON to enable

Step 5: Test with Copilot

Example 1: Asset Inventory

@agent How many cloud assets do I have?

Copilot: You have 1,247 cloud assets in your inventory.

Example 2: Category Search

@agent Show me 5 Container assets

Copilot: Here are 5 Container assets:
1. nginx-prod (Container) - Region: us-east-1
2. redis-cache (Container) - Region: us-west-2
3. postgres-db (Container) - Region: eu-west-1
4. kafka-broker (Container) - Region: us-east-1
5. redis-sentinel (Container) - Region: ap-south-1

Example 3: Security Analysis

@agent What security vulnerabilities do my AI models have?

Copilot: Your AI models have 147 security issues:
• ML Models: 5 issues (1 Critical, 4 Medium)
• LLM Models: 136 issues (136 Critical)
• Datasets: 6 issues (6 High)

Critical attention required for 137 vulnerabilities.

Example 4: Cloud Provider Filter

@agent List AWS assets in us-east-1 region

Copilot: Found 234 AWS assets in us-east-1 region...

Quick Reference

Server Commands

Action	Command
stdio server	`python3 MCP_server.py`
HTTP server	`python3 MCP_server_http.py`
stdio client	`python3 clients/stdio_client.py`
HTTP client	`python3 clients/http_client.py`

Configuration Paths

Tool	Config Location
VSCode Copilot	`.vscode/mcp.json` or `~/.vscode/mcp.json`
Gemini CLI	Managed by `gemini mcp` commands

AccuKnox MCP Server

AccuKnox MCP Server - Testing Guide

📋 Table of Contents

Prerequisites

Project Structure

Installation

1. Clone/Setup Project

2. Create Virtual Environment

3. Install Dependencies

4. Configure Environment

Testing stdio Server

With Test Client

With Gemini CLI

1. Install Gemini CLI

2. Configure MCP Server

3. Verify Configuration

4. Start Gemini CLI

5. Test with Natural Language Queries

6. Remove Server (Optional)

With VSCode Copilot (stdio)

1. Create Configuration File

2. Add stdio Server Configuration

3. Activate in VSCode

4. Test with Copilot

Testing HTTP Server

With HTTP Test Client

With VSCode Copilot (HTTP)

Prerequisites

Step 1: Verify HTTP Server

Step 2: Configure VSCode

Step 3: Activate Server

Step 4: Enable Copilot Agent Mode

Step 5: Test with Copilot

Quick Reference

Server Commands

Configuration Paths

Resources

Looking for Admin?

Latest Blog Posts

MCP directory API