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(8554)

0
0
High Cost
krajasek icon
Sec8

Automatically generate MCP (Model Context Protocol) servers from Python modules or installed packages, with LLM-powered documentation.

Setup Requirements

  • ⚠️Requires an LLM (Ollama, OpenAI, or Anthropic) for automatic tool descriptions unless `--no-llm` is used. Ollama requires a local server and model download. Cloud LLMs require an API key and internet access.
  • ⚠️For analyzing uninstalled Python packages, `uvx` (from `uv`) must be installed and available in PATH.
  • ⚠️Optional type adapters (e.g., pandas, Pillow, numpy) require those specific packages to be installed in the environment where `auto-mcp` runs or in the `uvx` isolated environment.
Verified SafeView Analysis
The project uses `subprocess.run`, `subprocess.Popen`, and `os.execvp` for its core `uvx`-based package isolation feature. This is a security-conscious design choice to run external package analysis in a sandboxed environment, rather than a vulnerability. The LLM is used for generating descriptions, not code, limiting direct LLM-related risks. API keys are handled via environment variables. The tool's primary function is to expose existing Python code; therefore, if a user exposes malicious local code or a compromised installed package, the generated server will inherit that risk. However, the tool itself does not introduce arbitrary `eval` or command injection vulnerabilities from external user input.
Updated: 2026-01-19GitHub→
0
0
Low Cost
dennisonbertram icon

mcp-nano-banana

by dennisonbertram

Sec9

Provides a Model Context Protocol (MCP) server for AI assistants like Claude to generate and manage images using Google's Gemini Nano Banana model.

Setup Requirements

  • ⚠️Requires a Google Gemini API key (image generation is a paid service).
  • ⚠️Requires Node.js >= 20.0.0 or Bun >= 1.1.0.
  • ⚠️Requires an active internet connection to communicate with the Google Gemini API.
Verified SafeView Analysis
The API key (GEMINI_API_KEY) is loaded from environment variables or a .env file, not hardcoded, which is a good practice. File system operations (saving images) are constrained to absolute paths and enforce a '.png' extension, mitigating common path traversal and arbitrary file write vulnerabilities. There is no apparent use of 'eval' or direct execution of arbitrary user-provided code. Communication is primarily via STDIO, reducing direct network exposure to external clients. Error handling for input validation (Zod) is present.
Updated: 2025-11-28GitHub→
0
0
Low Cost
Sarang-Nambiar icon

telegram-mcp

by Sarang-Nambiar

Sec7

Enables AI assistants to programmatically interact with Telegram chats for message reading, sending, and management.

Setup Requirements

  • ⚠️Requires an active Telegram account and API credentials (App ID and Hash).
  • ⚠️Requires `uv` package manager for setup and execution.
  • ⚠️A local session file (`anon.session`) is created upon first login and must be secured to prevent unauthorized Telegram account access.
Verified SafeView Analysis
The server loads Telegram API credentials from environment variables, which is good practice. It utilizes the Telethon library for API interaction. There are no explicit uses of `eval`, `exec`, or `os.system` for arbitrary code execution from user input found in the provided code. A critical security aspect is the `anon.session` file created locally, which stores session data; its compromise would grant access to the Telegram account. The `message` parameter in `send_message` is sent directly to Telegram, so an AI generating malicious markdown could potentially create annoying messages, but not execute code on the server itself. Overall, the direct code appears reasonably safe, but relies heavily on the security of the Telethon library and proper protection of the session file.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
lasaths icon
Sec8

This server provides a Model Context Protocol interface to access, search, and customize Phosphor Icons, offering various weights, colors, sizes, and implementation guidance for frontend development.

Setup Requirements

  • ⚠️Requires Node.js (version >=18.0.0)
  • ⚠️Dependencies must be installed via npm (npm install)
  • ⚠️Requires `npm run build` to compile TypeScript before running in production
Verified SafeView Analysis
The server fetches icons from a trusted GitHub repository and uses `zod` for input validation, which is good. Image processing is handled by `sharp`. The `saveToFile` and `saveDir` parameters in `get-icon` allow writing files to the local filesystem. While `path.resolve` and `path.dirname` are used to mitigate path traversal, and icon names are sanitized, any direct filesystem write from user input carries inherent risk if not perfectly secured. However, given the sanitization and path resolution, the immediate risk is low. No `eval` or blatant code injection vulnerabilities were found.
Updated: 2025-11-21GitHub→
0
0
Medium Cost
Kanishk31263 icon

Mcp-server-tools

by Kanishk31263

Sec9

Generates professional PPTX presentations from markdown content, leveraging an LLM for content creation.

Setup Requirements

  • ⚠️Requires PUPPER_AI_TOKEN environment variable (Puter AI account needed for LLM interaction)
  • ⚠️Requires Node.js 18 or higher
  • ⚠️Requires `assets/uaf_logo.png` image file to be present relative to the config for logo embedding
Verified SafeView Analysis
The server uses standard Node.js file system operations and `pptxgenjs` for presentation generation. File paths for output are controlled and relative to the server's directory, mitigating arbitrary file writes. It leverages `multer` for file uploads in `mcp-server-3.js`, a common but generally safe library, with output parsing via `pdf-parse`. The primary security consideration outside of standard web server practices is the reliance on an external LLM (Puter AI) for markdown generation, requiring trust in its output to prevent the generation of malicious content that might exploit the presentation generator (though no obvious vulnerabilities were found in the generator itself). No use of `eval` or suspicious obfuscation was detected.
Updated: 2025-12-09GitHub→
0
0
Medium Cost
zampierilucas icon

lkml-mcp

by zampierilucas

Sec9

Accesses and analyzes Linux Kernel Mailing List (LKML) threads and patches from lore.kernel.org.

Setup Requirements

  • ⚠️Requires Python 3.8+.
  • ⚠️Requires 'uvx' to run as an MCP server (install with `pip install uv`).
  • ⚠️Requires internet access to lore.kernel.org (or a configured `LKML_BASE_URL`).
Verified SafeView Analysis
The server makes HTTP requests to lore.kernel.org using the `requests` library. Message IDs are sanitized by stripping angle brackets before being used in URLs and as parts of temporary filenames, mitigating direct path traversal or URL injection. Temporary `.diff` files are created in `/tmp/lkml-mcp`, derived from email content. While the server itself doesn't execute these files, large or malformed content in `diff_text` could theoretically cause issues for external tools processing these files, but this is an indirect risk. No 'eval', 'exec', or direct shell execution with user input is observed. The base URL is configurable via environment variable, allowing a user to point to an untrusted source, but the default `lore.kernel.org` is secure.
Updated: 2025-12-01GitHub→
0
0
Low Cost

illustrator-mcp-server

by LatitudeMaster

Sec1

Serves as a backend component for Adobe Illustrator, likely managing extensions, plugins, or specific application features.

Review RequiredView Analysis
Source code was not provided for analysis. Therefore, a comprehensive security audit cannot be performed. It's impossible to verify the absence of 'eval' statements, obfuscation, network vulnerabilities, hardcoded secrets, or other malicious patterns without access to the code. Running this server without a code review is highly risky.
Updated: 2025-11-30GitHub→
0
0
Low Cost
pierrelouistheron icon

mcp-server-template-nodejs-19

by pierrelouistheron

Sec9

A TypeScript template for building Model Context Protocol (MCP) servers to enable communication with AI assistants and other MCP clients.

Setup Requirements

  • ⚠️Requires Node.js 22+
  • ⚠️A .env file needs to be created from .env.example, though `MCP_HTTP_PORT` has a default value.
Verified SafeView Analysis
The server uses Express.js and the Model Context Protocol SDK. Input validation for tools and prompts is handled via Zod schemas, which is a good practice to prevent common injection vulnerabilities. Environment variables are loaded using `dotenv`, promoting separation of configuration. No hardcoded secrets, direct `eval` of user input, or obvious insecure patterns were found in the provided source code. The main `/mcp` endpoint explicitly disallows GET and DELETE requests, focusing on POST for protocol communication. This template provides a good, secure starting point, but custom logic added by developers will require careful security review.
Updated: 2025-12-02GitHub→
0
0
Medium Cost
ivanmandat icon

fore-mcp-server

by ivanmandat

Sec7

This server provides a web-based interface and real-time communication layer to monitor, manage, and interact with a running Minecraft server using RCON.

Setup Requirements

  • ⚠️Requires a separate, running Minecraft Server to connect to.
  • ⚠️The Minecraft Server must have RCON enabled and configured with a password.
  • ⚠️Requires a Python environment with Flask, Flask-SocketIO, and other dependencies installed via pip.
Verified SafeView Analysis
The server exposes a web interface (Flask, SocketIO) and communicates with a Minecraft server via RCON. RCON connections are privileged and require strong password management, ideally loaded from environment variables via `config.py`. Standard web application security practices (authentication, authorization, input sanitization) are critical. No `eval` or obvious malicious patterns were found in the provided summaries. The security score reflects the inherent risks of managing RCON credentials and running a web service.
Updated: 2025-12-01GitHub→
0
0
Low Cost
Sec8

A complete Model Context Protocol (MCP) server demonstrating modern web development practices with authentication, billing, and PostgreSQL database integration for a Todo Management System, primarily interacted with via AI clients like Cursor AI.

Setup Requirements

  • ⚠️Requires Node.js version 18 or higher.
  • ⚠️Requires a Neon Database account (free tier available).
  • ⚠️Requires a Kinde account (free tier available).
Verified SafeView Analysis
The project integrates with external services (Kinde for OAuth, Neon for PostgreSQL) and uses JWT tokens and secure cookies for session management. It's a standard Node.js/TypeScript application; no signs of 'eval' or obfuscation were found. Risks are common to any web server handling user data and depend on proper configuration and secure coding practices within the application.
Updated: 2025-11-23GitHub→
0
0
Low Cost
burakarslan0110 icon

First-MCP-Server-Project

by burakarslan0110

Sec7

Provides a set of mathematical and string processing tools to AI assistants via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires .NET 10.0 SDK to be installed.
  • ⚠️Requires manual specification of the project path ('PROJE_YOLU') in the AI assistant's configuration.
Verified SafeView Analysis
The provided source code is truncated to only the README.md, making a full security audit of the C# implementation (Program.cs and Tools.cs) impossible. Based on the described functionality (basic math and string operations), the tools are not inherently risky, but potential vulnerabilities would depend on the actual C# code. The Factorial function explicitly limits input to 0-20, which is a good practice to prevent excessive computation or overflow.
Updated: 2025-12-13GitHub→
0
0
Low Cost
GaneshGMgr icon
Sec9

Manage personal income and expenses, including adding, listing, summarizing, editing, and soft-deleting financial entries.

Setup Requirements

  • ⚠️Requires Python 3.13 or newer.
  • ⚠️Requires 'uv' package manager for setup and running.
  • ⚠️On Windows, the 'fastmcp dev' CLI may fail, requiring direct Python execution with 'uv run python main.py'.
  • ⚠️Requires activation of the Python virtual environment (.venv) before running 'uv' commands.
Verified SafeView Analysis
The code uses parameterized queries for all database operations, effectively preventing SQL injection vulnerabilities. No 'eval' or similar dangerous functions were found. Input validation for dates and amounts is present. The server defaults to listening on '0.0.0.0', which is standard but implies it will be accessible from all network interfaces if deployed without appropriate firewall rules.
Updated: 2025-12-12GitHub→
PreviousPage 667 of 713Next