Stop Searching. Start Trusting.

The curated directory of MCP servers, vetted for security, efficiency, and quality.

Browse DirectoryWhy Us?

Tired of the MCP "Marketplace" Chaos?

We built MCPScout.ai to solve the ecosystems biggest pain points.

No Insecure Dumps

We manually analyze every server for basic security flaws.

Easy Setup

Our gotcha notes warn you about complex setups.

Avoid "Token Hogs"

We estimate token costs for cost-effective agents.

Products, Not Demos

We filter out "Hello World" demos.

CATEGORIES:
SORT:

Vetted Servers(9120)

0
0
Medium Cost
rmrfslashbin icon

manuals-mcp

by rmrfslashbin

Sec7

An MCP server to access and manage hardware and software documentation, including device details, pinouts, specifications, and full-text search, by connecting to a Manuals REST API backend.

Setup Requirements

  • ⚠️Requires a running instance of the Manuals REST API to connect to.
  • ⚠️Requires a Manuals API Key for full read/write and administrative features (read-only is anonymous).
  • ⚠️The `upload_file` and `publish` tools can read arbitrary files from the local filesystem based on AI input, which may pose a security risk for information disclosure.
Review RequiredView Analysis
The server correctly uses environment variables or a configuration file for sensitive API keys, avoiding hardcoding. It acts as an API client, not exposing direct network ports for external interaction. However, the `upload_file` and `publish` tools accept a `local_path` argument from the AI model to read files from the local filesystem. This could pose an information disclosure risk if a malicious or compromised AI requests sensitive local files that the MCP server process has permission to read (e.g., `/etc/passwd`, user configuration files) without any explicit path sanitization or restrictions in the provided code. The server itself does not contain `eval` or `os/exec` patterns that would directly lead to arbitrary code execution based on AI input, making the primary risk information exfiltration via `local_path` if an untrusted AI is used.
Updated: 2025-12-14GitHub→
0
0
Low Cost

kali-linux-mcp

by JesseEikeland

Sec4

Provides a Claude Desktop-integrated MCP server for accessing over 30 Kali Linux penetration testing tools.

Setup Requirements

  • ⚠️Requires Docker Desktop to be installed.
  • ⚠️Requires Claude Desktop to be installed.
  • ⚠️Needs at least 8GB RAM and 20GB free disk space.
  • ⚠️Initial Docker container build takes 10-15 minutes.
  • ⚠️Requires manual configuration of Claude Desktop's `claude_desktop_config.json` with a specific, path-dependent entry.
Review RequiredView Analysis
The server runs Kali Linux tools in a Docker container configured with `network_mode: host` and `NET_ADMIN`, `NET_RAW` capabilities. While the application implements some internal sanitization, command timeouts, and non-root execution, these container capabilities grant direct, unfiltered access to the host's network stack, which is a significant security risk, especially given the offensive nature of the tools. This setup is inherently dangerous if not used in a strictly controlled and isolated environment.
Updated: 2025-11-17GitHub→
0
0
Low Cost

mcp-server

by SBourrier

Sec5

A C# .NET 8 MCP server that allows LibreChat to retrieve SQL Server version information via a provided connection string.

Setup Requirements

  • ⚠️Requires .NET 8 SDK or Docker runtime.
  • ⚠️Requires a running SQL Server instance accessible from the MCP server (e.g., using `host.docker.internal` from Docker on Windows/Mac).
  • ⚠️LibreChat integration requires specific YAML configuration and restarting LibreChat.
Review RequiredView Analysis
The `tools/call` method accepts a `connectionString` as an argument, allowing users to specify arbitrary SQL Server connection details. This poses a significant risk for unauthorized database access or data exfiltration if the MCP server is exposed to untrusted clients or networks, or if users within LibreChat are not fully trusted. Using `TrustServerCertificate=True` further weakens security for SQL connections. It should only be run in highly controlled and trusted environments with strict network isolation and access controls.
Updated: 2025-11-17GitHub→
0
0
High Cost
NoCodeify icon

aeo

by NoCodeify

Sec9

Performs Answer Engine Optimization (AEO) audits across ChatGPT, Gemini, and Google Search, comparing LLM responses, extracting citations, and generating comprehensive brand audit reports with detailed analysis.

Setup Requirements

  • ⚠️Requires paid API keys for OpenAI, Google GenAI, and ScrapingBee.
  • ⚠️Requires manual configuration in the Claude Code config file (~/.claude/claude_desktop_config.json).
  • ⚠️Calls to 'run_brand_audit' execute 8 queries across 3 engines (ChatGPT, Gemini, Google Search), incurring significant costs due to multiple LLM inferences and web scrapes.
Verified SafeView Analysis
API keys are properly handled via environment variables. Input parameters for tools are validated using Zod, mitigating some injection risks to the external APIs. Network calls are made to trusted external APIs (OpenAI, Google GenAI, ScrapingBee). No critical security vulnerabilities like 'eval' or direct arbitrary file system access are identified in the provided truncated code.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
aplaceforallmystuff icon

mcp-wp-abilities

by aplaceforallmystuff

Sec9

Exposes WordPress 6.9+ Abilities API as AI-accessible tools for dynamic interaction with a WordPress site.

Setup Requirements

  • ⚠️Requires Node.js 18+
  • ⚠️Requires WordPress 6.9+ with the Abilities API enabled
  • ⚠️Requires a WordPress Application Password for authentication
Verified SafeView Analysis
The server relies on WordPress Application Passwords for authentication, which grants full API access for the associated user. The README correctly advises using a dedicated user with minimal required capabilities and storing credentials securely via environment variables. No 'eval' or obvious code injection vulnerabilities were found. Input validation is delegated to the WordPress Abilities API itself, which is a standard practice.
Updated: 2026-01-16GitHub→
0
0
Low Cost
DareDev256 icon

fcp-mcp-server

by DareDev256

Sec9

This server enables AI assistants to interact with and control Final Cut Pro projects through natural language by parsing and modifying FCPXML files.

Setup Requirements

  • ⚠️Requires Python 3.10+
  • ⚠️Requires an MCP-compatible AI client (e.g., Claude Desktop)
  • ⚠️Requires Final Cut Pro 10.4+ (for FCPXML compatibility)
Verified SafeView Analysis
The server primarily operates on local FCPXML files, limiting external attack surfaces. It uses standard Python libraries for XML parsing and file system interaction (pathlib), without obvious signs of direct shell command execution, 'eval' or 'exec', or hardcoded secrets. File paths are derived from an environment variable or directly passed as tool arguments, implying that the AI client is a trusted orchestrator. The main risk would stem from malicious FCPXML files themselves, which are assumed to be sourced by the user.
Updated: 2026-01-18GitHub→
0
0
High Cost
UzmaFateh icon

mcp-servers

by UzmaFateh

Sec5

Manages and runs Minecraft servers, likely with custom modifications or enhanced features given the 'MCP' (Mod Coder Pack) context.

Setup Requirements

  • ⚠️Requires a Java Runtime Environment (JRE) installed
  • ⚠️Can be highly demanding on CPU and RAM resources
  • ⚠️May require port forwarding for external multiplayer access
Review RequiredView Analysis
A thorough security audit could not be performed as no source code was provided for analysis. Without access to the code, it's impossible to check for dangerous patterns like 'eval', obfuscation, hardcoded secrets, network risks, or malicious logic. The score is a neutral placeholder reflecting this lack of information.
Updated: 2025-12-05GitHub→
0
0
Medium Cost
ankitk090-lang icon

Zantrix

by ankitk090-lang

Sec7

Provides a suite of cybersecurity diagnostic and network analysis tools that integrate with MCP clients like Claude Desktop.

Setup Requirements

  • ⚠️Requires Node.js (v16 or higher) and npm.
  • ⚠️Requires an MCP client (e.g., Claude Desktop) and manual configuration of its `mcpServers` config file, including providing the absolute path to the server's `dist/index.js` file.
Verified SafeView Analysis
The server's core functionality involves performing network requests to user-provided IPs/domains/URLs (e.g., trace_ip, scan_network, analyze_headers, dns_lookup, whois_lookup, ssl_inspector, find_subdomains, check_robots) and accessing local file paths (e.g., scan_files, hash_file, search_logs). While these are the intended security features, they inherently grant the integrated AI powerful capabilities for local file system interaction and external/internal network reconnaissance from the user's machine. This means a malicious or compromised AI, or a user tricked by a prompt, could potentially misuse these tools to read sensitive local files (e.g., scan_files, search_logs) or perform network activities. No 'eval' or direct self-executing malicious patterns are found in the server's own code; 'eval' is used as a *pattern to detect* in `scanFiles.ts`. No hardcoded secrets were identified. The `@ts-ignore` for the 'whois' library is a minor code quality note but not a direct security vulnerability.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
mnemoverse icon

mcp-docs-server

by mnemoverse

Sec3

A universal template for creating Model Context Protocol (MCP) servers that provide AI agents with documentation access.

Setup Requirements

  • ⚠️Requires Node.js 18+.
  • ⚠️The default documentation (`docs/`) needs to be replaced with your own markdown files for custom content.
  • ⚠️CRITICAL SECURITY VULNERABILITY: Path Traversal in `get_document` tool allows arbitrary file reads. Not safe to run without patching input validation for document paths.
Review RequiredView Analysis
CRITICAL Path Traversal Vulnerability: The `getDocument` function in `src/lib/DocumentManager.ts` uses `path.join(this.docsPath, documentPath)` to resolve document paths. If `documentPath` is provided by an AI agent (or other user input) as an absolute path (e.g., `/etc/passwd` or `C:/Windows/System32/drivers/etc/hosts`), `path.join` will resolve to this absolute path, ignoring the intended `docsPath`. This allows arbitrary file reading outside the designated documentation directory. The server also executes shell commands in its build/template scripts but these are not runtime risks.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
mem-kit icon

camunda-mcp

by mem-kit

Sec8

Provides Model Context Protocol (MCP) services for interacting with Camunda BPM Platform, a simulated StateStreet Customer Due Diligence (CDD) service, and an OCR service via Tencent Cloud.

Setup Requirements

  • ⚠️Requires a running Camunda BPM Platform instance (e.g., via Docker).
  • ⚠️Requires Tencent Cloud API credentials for the OCR service (may incur costs).
  • ⚠️Environment variables (CAMUNDA_URL, CAMUNDA_USER, CAMUNDA_PASSWORD, TENCENTCLOUD_SECRET_ID, TENCENTCLOUD_SECRET_KEY) must be configured in a .env file.
  • ⚠️Python virtual environment setup and `pip install -r requirements.txt` are required for each service.
Verified SafeView Analysis
Secrets (Camunda credentials, Tencent Cloud API keys) are correctly loaded from environment variables. No 'eval' or malicious obfuscation found. The `deploy_bpmn` tool allows deploying files to Camunda, which is a powerful operation and should only be used by trusted clients. The `invoice_ocr` tool forwards image URLs directly to Tencent Cloud, which might pose SSRF risks if the Tencent API does not validate URLs, but this is an inherent function of the tool. Broad CORS (`allow_origins=["*"]`) is enabled for all services, which might be too permissive for production environments without specific security considerations. Overall, the server's operations are transparent and intended.
Updated: 2025-11-29GitHub→
0
0
Medium Cost
agentika icon

mcp-servers

by agentika

Sec3

This project provides a multi-agent AI bot that integrates various specialized 'MCP Servers' (Multi-Agent Communication Protocol) into a unified interface, allowing the main agent to leverage different tools and capabilities provided by sub-agents for tasks like note-taking, planning, and arithmetic.

Setup Requirements

  • ⚠️Requires 'uv' for installation and execution.
  • ⚠️Requires API keys for a large language model provider (OpenAI, Azure, or Gemini), which are paid services.
  • ⚠️Requires Python 3.12 or newer.
  • ⚠️The 'notemcp' server, included in the default configuration, requires the `NOTEMCP_BASE_DIR` environment variable to be set for note storage.
Review RequiredView Analysis
The core application loads MCP server configurations from JSON files (e.g., `config/default.json`), which specify `command` and `args` to execute external processes. This design choice allows for arbitrary command execution on the host system if the configuration file is sourced from an untrusted location or compromised. There is no explicit sandboxing or validation of these commands within the provided code, posing a significant security risk for arbitrary code execution if config integrity is not guaranteed. Hardcoded secrets or obvious 'eval' usage were not found in the main application logic, but the inherent command execution capability is critical.
Updated: 2025-11-20GitHub→
0
0
Medium Cost
Beerspitnight icon
Sec9

Enables Claude Code CLI or Desktop to interact with Spotify for advanced playlist curation, management, and audio feature analysis.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Requires a Spotify Developer App for API credentials (Client ID, Client Secret, Redirect URI).
  • ⚠️Requires a one-time browser authentication step to generate a `.spotify_cache` token.
  • ⚠️GetSongBPM API Key is optional but recommended for enhanced audio feature coverage.
Verified SafeView Analysis
The project uses environment variables for API credentials and explicitly warns against committing `.env` and `.spotify_cache` (which contains tokens) to version control. API calls include retry logic and timeouts, which are good practices. No obvious malicious patterns or unsafe code execution (`eval`) were found in the provided source.
Updated: 2025-11-28GitHub→
PreviousPage 285 of 760Next