MCPS is now an IETF Internet-Draft.Read the spec
Free & Open Source OWASP MCP Top 10 OWASP Agentic AI Top 10 PDF Reports

Security Scanner for MCP Servers & AI Agents

One command. 22 security checks. Professional PDF report. Scan your MCP server or AI agent against OWASP MCP Top 10 + OWASP Agentic AI Top 10.

# Install and scan in 30 seconds
npx mcps-audit ./my-mcp-server
open mcps-audit-report.pdf
Install from npm View Sample Report GitHub

How It Works

You have code. You want to know if it's safe. mcps-audit reads your code files, looks for 22 types of security problems across two OWASP standards, and tells you:

PASS

No serious issues. Your agent or MCP server looks good.

⚠️

WARN

Some problems found. You should fix these before shipping.

FAIL

Critical issues. Fix these immediately before deploying.

It shows you which file, which line, and how to fix it. The PDF report is designed to hand to your security team, compliance officer, or client.

1. OWASP MCP Top 10

10 security risks specific to Model Context Protocol servers. The scanner checks for vulnerable patterns AND whether mitigations (like MCPS) are in place.

IDRiskWhat It Means
MCP-01Rug PullsTool definitions change after approval
MCP-02Tool PoisoningHidden instructions in tool descriptions
MCP-03Privilege EscalationCombining tools to get more access than intended
MCP-04Cross-Server ForgeryOne MCP server tricks another
MCP-05Sampling ManipulationServer manipulates AI responses
MCP-06Prompt Injection via MCPMalicious data injected through tool responses
MCP-07Resource ExhaustionNo auth means anyone can abuse your server
MCP-08Insufficient LoggingNo audit trail for what happened
MCP-09Insecure MCP-to-MCPNo origin validation between servers
MCP-10Context PollutionMalicious data pollutes shared context

2. OWASP Agentic AI Top 10

12 security rules for AI agents mapped to MITRE ATT&CK techniques and STRIDE threat model.

RuleChecks ForSeverity
AS-001exec(), eval(), subprocess -- dangerous code executionCRITICAL
AS-002Hardcoded API keys, passwords, tokensHIGH
AS-003Excessive permissions (admin, delete, execute)MEDIUM
AS-004File input flowing into prompts (injection vector)HIGH
AS-005Known injection: SQL, XSS, command injectionCRITICAL
AS-006Code execution without sandboxingHIGH
AS-007Dependencies without lockfile or integrityLOW
AS-008Auto-approve, bypass safety, skip confirmationHIGH
AS-009innerHTML, document.write -- unsafe outputMEDIUM
AS-010No logging or monitoring detectedMEDIUM
AS-011HTTP requests that could exfiltrate dataHIGH
AS-012Server endpoints without authenticationHIGH

Tested Against Real Frameworks

We tested mcps-audit against major open-source agent frameworks and official MCP servers. Zero crashes across 1,700+ files and 450,000+ lines of code.

CrewAI
Agent Framework
FAIL
834 findings across 500 files
LangGraph
Agent Framework
FAIL
1,018 findings across 322 files
Pydantic AI
Agent Framework
FAIL
772 findings across 435 files
MCP Filesystem
Official MCP Server
FAIL
28 findings across 15 files
MCP Fetch
Official MCP Server
WARN
7 findings across 4 files
MCP Memory
Official MCP Server
PASS
23 findings across 6 files

All targets are public open-source repositories (MIT/Apache 2.0). Findings represent pattern-based static analysis, not confirmed vulnerabilities.

Part of the MCPS Ecosystem

mcps-audit is the testing tool that complements the MCPS standard and SDK. Together they form a complete security lifecycle for MCP.

Standard
MCPS IETF Draft
SDK
mcp-secure
Tester
mcps-audit
Platform
AgentSign

Scan first. Fix what mcps-audit finds. Then add mcp-secure to protect against the MCP Top 10.

IETF Spec npm: mcp-secure OWASP Alignment

Usage

# Scan current directory
npx mcps-audit
# Scan a specific folder, name it in the report
npx mcps-audit ./src --name "My MCP Server"
# Only HIGH and CRITICAL issues
npx mcps-audit . --severity HIGH
# Custom output path
npx mcps-audit ./server -o ~/Desktop/security-report.pdf
# JSON output for CI/CD
npx mcps-audit . --json
# GitHub Actions
- run: npx mcps-audit . --severity HIGH
# Exit code 1 if FAIL, 0 if PASS/WARN
OptionWhat It DoesDefault
targetFolder or file to scan. (current directory)
-o, --outputWhere to save the PDF./mcps-audit-report.pdf
--nameName shown in the reportFolder name
--jsonPrint JSON to terminalOff
--severityMinimum level: LOW, MEDIUM, HIGH, CRITICALLOW

The PDF Report

The generated report includes 8 sections designed for security teams, compliance audits, and executive review:

1. Cover Page

Target name, date, report ID, scanner version

2. Executive Summary

PASS/WARN/FAIL verdict, risk score, severity breakdown

3. Risk Comparison

"WITHOUT MCPS" vs "WITH MCPS" side-by-side with reduction %

4. MCP Top 10 Matrix

Pass/Fail/Warn for each of the 10 MCP risks

5. Agentic AI Matrix

12 rules with MITRE ATT&CK + STRIDE mapping

6. Detailed Findings

File path, line number, code snippet, how to fix

7. Remediation Checklist

Prioritized by severity, checkbox format

8. Methodology

Standards referenced, scanner version, contact

Download Sample Report (PDF)

Requirements

Node.js 18+

Only requirement. Works on macOS, Linux, Windows.

📦

One Dependency

pdfkit for PDF generation. No Chrome, no Puppeteer.

🚀

Instant

Scans 500 files in under 2 seconds. No cloud, no API keys.