# MCP Trust Registry

> The MCP Trust Registry is a security-focused registry that scans MCP (Model Context Protocol) server builds for vulnerabilities. It provides risk ratings, code-level evidence, and remediation guidance — mapped to OWASP MCP Top 10, MAESTRO, and MITRE CWE frameworks. Built and maintained by BlueRock.

## What It Does

The MCP Trust Registry analyzes MCP server source code across 22 security rules and returns findings with evidence down to the line of code, plus remediation guidance. It scans both public MCP servers and private/enterprise builds.

As of early 2026, the registry has scanned 10,000+ public MCP servers. Key findings:

- 9.2% have critical vulnerabilities
- 36.7% have unbounded URI / SSRF vulnerabilities
- 43% have command injection flaws

New MCP servers are being published at a rate of 20,000+ per month.

## What It Scans For

The registry analyzes MCP server builds across five vulnerability categories:

| Category | What It Catches |
|---|---|
| Unrestricted network fetch + SSRF | Unbounded URI calls, metadata service access, credential exfiltration |
| Command injection | Process execution without input sanitization, shell spawning |
| Denial of wallet / cost controls | Unthrottled API calls, missing rate limits, resource exhaustion |
| Dependency risk + supply chain | Vulnerable dependencies, unvetted packages, version pinning gaps |
| Transport layer security | Missing TLS, insecure connections, cleartext transmission |

All 22 rules are mapped to OWASP MCP Top 10, MAESTRO, and MITRE CWE frameworks.

## How to Use It

### Browse Public Servers
Search 10,000+ public MCP server scans at https://mcp-trust.com. See risk ratings, vulnerability details, and remediation guidance before connecting a server to your agent.

### Scan Your Private Build
Organizations can submit internal MCP server repositories for the same 22-rule analysis — same rules, same evidence depth, same remediation guidance. Available for enterprise and security teams.

## Who It's For

- **AI agent developers** checking MCP server safety before connecting
- **MCP server builders** validating their own implementations
- **Product security teams** evaluating MCP servers during approval processes
- **DevOps teams** integrating security checks into CI/CD pipelines

## Notable Research

- **Markitdown SSRF finding (February 2026):** BlueRock research revealed an unbounded URI call in Microsoft's Markitdown MCP server (86,000+ GitHub stars) that could access cloud instance metadata credentials. Registry analysis found 36.7% of scanned servers share the same vulnerability class.

## Key Data

- 10,000+ public MCP servers scanned
- 9.2% have critical vulnerabilities
- 36.7% have unbounded URI / SSRF vulnerability
- 43% have command injection flaws
- 22 security rules mapped to OWASP MCP Top 10, MAESTRO, and MITRE CWE
- 20,000+ new MCP servers published monthly
- Code-level findings with evidence down to the line

## About BlueRock

BlueRock provides observability, context, and control for agentic AI systems operating in production — powered by the Trust Context Engine. The MCP Trust Registry is one of three BlueRock products. Teams use it to know what's safe before they connect.

## Links

- MCP Trust Registry: https://mcp-trust.com
- BlueRock: https://bluerock.io
- Private scanning: https://bluerock.io/mcp-trust-registry