ScannerMIT·Python

Sentinel

Automated security scanner for MCP server infrastructure.

MCP servers ship with zero security defaults. No authentication required, no input validation enforced, no monitoring configured. Sentinel runs 26 detection rules across three modules to surface those gaps before an attacker finds them, and plugs directly into your CI/CD pipeline.

View on GitHub

Status

Active

Language

Python

Rules

26 detection rules

License

MIT

Install

pip install helixar-sentinel

The Problem

Model Context Protocol servers are typically deployed from framework quickstarts and documentation examples that optimise for getting something working, not for getting something secure. The result is consistent: unauthenticated endpoints, tools exposed with no permission boundaries, containers running as root, no secrets rotation, no egress filtering.

Manual security reviews catch some of these issues but do not scale. Code review misses runtime misconfigurations. DAST tools are not tuned for MCP server behaviour. No dedicated scanner existed for this class of infrastructure until Sentinel.

Sentinel closes that gap with a purpose-built scanner that understands MCP server architecture: 26 rules across static config, live endpoint probing, and container inspection. Drop it into CI/CD and get severity-rated findings with remediation guidance on every run.

Three Scanning Modules

10 rules

Configuration Analysis

Static analysis of MCP server configuration files. Surfaces missing authentication, overly broad tool permissions, insecure defaults, and exposed secrets in config.

8 rules

Live Endpoint Assessment

Active probing of running MCP server endpoints. Detects unauthenticated access, missing TLS, CORS misconfigurations, and exposed debug interfaces.

8 rules

Container Inspection

Analyses container images and runtime configurations for MCP deployments. Flags privileged containers, missing resource limits, and insecure base images.

GitHub Code Scanning Integration

Sentinel outputs SARIF 2.1, the format natively understood by GitHub Code Scanning and most enterprise SAST platforms. Findings appear inline in pull requests, with configurable fail thresholds per severity level (critical → low). No extra tooling required.

Quick Start

Install from PyPI and run your first scan in under five minutes.

terminal

$ pip install helixar-sentinel

# Scan a running MCP server
$ sentinel scan --target http://localhost:3000 --config ./mcp-config.json
  [CRITICAL] No authentication on tool endpoint /tools/execute
  [CRITICAL] Container running as root (UID 0)
  [HIGH]     Missing rate limiting on /tools/execute
  [HIGH]     Debug interface exposed on :9229
  [PASS]     TLS enforced on all endpoints

# Output SARIF for GitHub Code Scanning
$ sentinel scan --output sarif --fail-on critical > results.sarif