owasp-mcp

Identifies, assesses, and remediates Model Context Protocol security risks using the OWASP MCP Top 10.

1.4k|284|Updated Nov 2, 2025
One-click install
npx skills add https://github.com/microsoft/hve-core --skill owasp-mcp
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: owasp-mcp
Source: https://github.com/microsoft/hve-core/tree/main/.github/skills/security/owasp-mcp
Command: npx skills add https://github.com/microsoft/hve-core --skill owasp-mcp

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Teams deploying Model Context Protocol servers and agents lack a structured way to identify and remediate MCP-specific security risks such as token leakage, tool poisoning, prompt injection, and shadow servers.

Core Features & Use Cases

  • Structured Vulnerability Knowledge Base: Encodes the OWASP MCP Top 10 (2025) as machine-readable reference documents covering MCP01 through MCP10.
  • Assessment Checklists: Each vulnerability includes impact analysis, vulnerability checklists, prevention controls, attack scenarios, detection signals, and remediation steps.
  • Use Case: When reviewing an MCP server deployment, query the skill to check for risks like privilege escalation via scope creep or insufficient authentication, then apply the listed prevention controls such as mTLS, scoped tokens, and audit logging.

Quick Start

Review my MCP server configuration against the OWASP MCP Top 10 and list any vulnerabilities with remediation steps.

Frequently Asked Questions about owasp-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What is the OWASP MCP Top 10?

The OWASP MCP Top 10 (2025) is a ranked list of the most critical security risks for Model Context Protocol systems, covering issues from token mismanagement (MCP01) to context injection and over-sharing (MCP10). Each entry includes impact, checklists, controls, and remediation guidance.

How do I assess an MCP server for security vulnerabilities?

Use the vulnerability checklists in each reference document to evaluate your MCP deployment against risks like hard-coded tokens, missing mTLS, unsigned schemas, and unregistered shadow servers. Each checklist maps to concrete prevention controls and detection signals.

How do I prevent prompt injection in MCP agents?

Treat retrieved context as untrusted data: sanitize metadata, enforce instruction precedence, track provenance, and require human approval for sensitive tool calls. Constrain model output to structured formats validated against schemas.

What are shadow MCP servers and why are they risky?

Shadow MCP servers are unapproved MCP deployments operating outside formal security governance, often with default credentials and permissive configurations. They bypass centralized authentication and monitoring, creating unmonitored endpoints vulnerable to exploitation and data exfiltration.

Does this skill detect vulnerabilities automatically?

No, it is a knowledge base of structured reference documents rather than a scanning tool. An agent queries the references to identify risks, apply checklists, and recommend remediation, but it does not execute scans or runtime detection itself.