skills.rest
Wyl-cmd
Wyl-cmdCOMMUNITY
Explore all skills by this author →
GitHub

hunt-mcp-security

Tests Model Context Protocol servers for tool access control, injection, and output poisoning vulnerabilities.

6|Updated Jan 31, 2026
Search Tags:#vulnerability assessment#penetration testing#mcp#red team#ai security#prompt injection#tool security
One-click install
npx skills add https://github.com/Wyl-cmd/kxns-cli --skill hunt-mcp-security-wyl-cmd
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: hunt-mcp-security
Source: https://github.com/Wyl-cmd/kxns-cli/tree/main/src/kxns_cli/skills/hunt-mcp-security
Command: npx skills add https://github.com/Wyl-cmd/kxns-cli --skill hunt-mcp-security-wyl-cmd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI agents connected to external tools via the Model Context Protocol (MCP) create a new attack surface where unauthenticated tool access, unsanitized inputs, and poisoned tool outputs can lead to privilege escalation and data exfiltration. This Skill provides a structured methodology to identify and verify these vulnerabilities before attackers do. ## Core Features & Use Cases - Tool Access Control Testing: Enumerate exposed MCP tools and check whether sensitive operations can be invoked without authentication or proper authorization. - Injection & Output Poisoning Probes: Test tool parameters for SQL injection, command injection, NoSQL injection, and prototype pollution, and verify whether crafted tool outputs can inject instructions back into the agent. - Use Case: During a security assessment of an application deploying Claude Desktop MCP or a custom agent framework, use this Skill to map the tool catalog via the /tools endpoint, then systematically test each tool for missing access controls, cross-user data isolation failures, and unsafe tool registration. ## Quick Start Use the hunt-mcp-security skill to test the MCP server at the target URL for unauthenticated tool access and injection vulnerabilities.

Frequently Asked Questions about hunt-mcp-security

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

FAQPage Schema
How do I test an MCP server for security vulnerabilities?▼

Start by enumerating the tool catalog through the /mcp/tools or /mcp/schema endpoints, then test each tool for unauthenticated access, injection through parameters, and excessive permissions. This Skill provides a seven-phase procedure covering access control, injection, output poisoning, and data isolation.

What is MCP tool output poisoning?▼

Tool output poisoning occurs when a malicious tool response is injected back into the AI agent's context, potentially overriding its instructions. Test it by having a fetch tool load a page containing hidden directives and observing whether the agent acts on them.

Does this MCP security testing skill work on Windows?▼

No, the skill requires a Linux environment with curl and python3, as declared in its platform and compatibility metadata. It is designed to run within Kali Linux-based penetration testing environments.

How do I check if MCP tools require authentication?▼

Send POST requests to tool execution endpoints without credentials and observe the HTTP status codes. The skill includes a loop that tests common sensitive tool names like delete_users, execute_sql, and reset_passwords for unauthenticated access.

What are the limitations of MCP vulnerability scanning?▼

MCP is a protocol standard, not a single implementation, so tool schemas and access patterns vary per server and must be mapped first. Also, unauthenticated access to intentionally public tools like weather or news APIs is not a valid finding.

Related Skills

Explore More Software Engineering Skills→
wgpsecwgpsec
1.6k

recon-full

Enumerate subdomains, scan ports, fingerprint assets, and validate POCs.

Official
Advanced
kaminocorpkaminocorp
1

Vulnerability Report Writing

Generate bounty-ready vulnerability reports with CVSS scoring and evidence-based severity calibration.

Official
Intermediate
jayjpatel9717jayjpatel9717

llm-security

Detect and mitigate LLM security vulnerabilities like prompt injection and RAG poisoning.

Community
Advanced
OrcaQubitsOrcaQubits
35

acp-checkout-mcp

Expose ACP checkout operations as callable MCP tools for AI agents.

Official
Intermediate