Agentic Security

Provides real-time AI security for agentic coding, including prompt firewall, hallucination shield, and vulnerability scanning via MCP.

javascript

6

GitHub Stars

javascript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "sinewaveai-agent-security-scanner-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "agent-security-scanner-mcp"
      ]
    }
  }
}

You can protect AI coding agents in real time with Agentic Security MCP Server, a security layer that analyzes agent instructions, verifies packages, and scans code for AI-specific vulnerabilities as you work. It integrates with your MCP-enabled tooling to guard prompts, validate dependencies, and surface auto-fix suggestions while you code.

How to use

Set up the MCP server in your development environment and connect your MCP-enabled AI coding tools. Once configured, you can run real-time checks as your agent receives prompts, analyzes code, and suggests or applies fixes for vulnerabilities. The three pillars—Prompt Firewall, Hallucination Shield, and the Vulnerability Scanner—work together to block malicious prompts, detect hallucinated packages, and surface concrete remediation steps.

How to install

Prerequisites: Node.js v18 or later. Ensure you have a functioning MCP-enabled client or editor that can connect to an MCP server.

npm install -g agent-security-scanner-mcp

Or run directly without installing:```bash npx agent-security-scanner-mcp

## Additional sections

Configuration is provided through MCP connections. For local usage, the server runs as a local process via a command like npx agent-security-scanner-mcp, and you connect your MCP client to it. No external URLs are required if you use the local endpoint.

Tools and endpoints described here include prompt analysis, package verification, and vulnerability scanning. You can enable auto-fix templates for common vulnerabilities and list all available security rules to tailor your workflow.

## Available tools

### scan\_agent\_prompt

Analyze prompts for malicious intent before execution to prevent prompt injection and data exfiltration.

### check\_package

Verify if a package exists in the official registry before use.

### scan\_packages

Scan files for potentially hallucinated or dangerous imports.

### list\_package\_stats

Show statistics about loaded package databases across ecosystems.

### scan\_security

Scan code for vulnerabilities with fix suggestions.

### fix\_security

Auto-apply all available security fixes.

### list\_security\_rules

List all security rules with metadata.