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
Medium Cost
Mainakcris7 icon

mcp-python-demo

by Mainakcris7

Sec4

Provides an MCP (Micro-Agent Communication Protocol) server exposing tools for basic arithmetic and fetching real-time, formatted weather information from external APIs.

Setup Requirements

  • ⚠️Requires Python 3.14+.
  • ⚠️Requires API keys for geocoding (geocode.maps.co) and OpenWeatherMap, stored as environment variables GEOCODING_API_KEY and OPEN_WEATHER_API_KEY.
  • ⚠️The server uses 'requests.get(..., verify=False)', which disables SSL verification and poses a security risk; consider removing 'verify=False' in production.
Review RequiredView Analysis
The use of 'requests.get(..., verify=False)' in the 'get_weather_data' function disables SSL certificate verification, making the application vulnerable to Man-in-the-Middle (MITM) attacks. While API keys are loaded from environment variables (good practice), this severe vulnerability significantly reduces the security posture. No other obvious critical flaws like 'eval' or hardcoded secrets were found.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
skribblez2718 icon

mcp-servers

by skribblez2718

Sec7

An MCP server for the Recipez recipe management API, providing tools for recipe CRUD, AI-powered generation, grocery lists, and email operations.

Setup Requirements

  • ⚠️Requires a separate, live instance of the 'Recipez API' with JWT authentication.
  • ⚠️Requires the 'uv' package manager for installation and running.
  • ⚠️Requires Python 3.10 or higher.
  • ⚠️The `RECIPEZ_JWT_TOKEN` needs to be actively managed and refreshed, as it expires every 12 hours (by default of the upstream Recipez API).
  • ⚠️Default `CORS_ORIGINS='*'` is insecure and must be configured for production.
Verified SafeView Analysis
The server demonstrates good input validation (e.g., UUID validation, specific string lengths/formats), structured logging, and robust error handling. Sensitive credentials (JWT token, base URL) are correctly designated as environment variables rather than hardcoded. However, the default `CORS_ORIGINS` is set to `*` which is a significant security risk in production environments as it allows any origin to access the API. This requires manual adjustment by the user for secure deployment. Additionally, the README indicates the `RECIPEZ_JWT_TOKEN` expires after 12 hours, implying a need for manual refreshing or an external token management system not explicitly detailed in the provided server code, which could lead to service interruptions or insecure handling if not properly addressed by the operator.
Updated: 2026-01-03GitHub→
0
0
Low Cost
ktdreyer icon

openai-mcp

by ktdreyer

Sec9

Manages OpenAI organization invitations by allowing an LLM to check invite status and send new invites.

Setup Requirements

  • ⚠️Requires two restricted OpenAI admin API keys (read and write scopes) due to an OpenAI limitation, which must be set as environment variables.
  • ⚠️Requires `uv` and `mcp[cli]` Python packages to be installed.
  • ⚠️For production use, the design's reliance on LLM-provided user emails for invitations should be addressed with a secure authentication mechanism.
Verified SafeView Analysis
The server code itself avoids common pitfalls like hardcoded secrets or `eval` by using environment variables for API keys. It leverages restricted OpenAI admin keys as recommended. A critical security note in the README indicates that user-provided email addresses to the LLM are a 'naive example' and in a production deployment, authenticated user email addresses should be passed through a secure mechanism to prevent trusting unverified LLM input. This is a design-level risk for a production system, not a direct vulnerability in the provided server code.
Updated: 2025-12-10GitHub→
0
0
High Cost
vijaypandiyan icon

VbMigrationAssistant

by vijaypandiyan

Sec5

Automated analysis, planning, and migration of VB.NET and C# projects (including Web Forms) to C# on .NET 9 using an MCP server and LLM assistance.

Setup Requirements

  • ⚠️Requires an Anthropic API Key (Paid) if 'LlmProvider' is 'claude'.
  • ⚠️Requires a local Ollama instance running if 'LlmProvider' is 'ollama'.
  • ⚠️Requires 'dotnet tool install upgrade-assistant --global' for framework upgrade functionality.
  • ⚠️Requires Git to be installed for repository cloning.
  • ⚠️PowerShell 7+ is required to run the provided client scripts.
Review RequiredView Analysis
The server's core functionality involves cloning arbitrary GitHub repositories from a user-provided 'repoUrl'. It then parses, analyzes, converts, and potentially builds these projects (e.g., 'dotnet build' is mentioned in the cleanup phase). Building or running tools like 'dotnet upgrade-assistant' on untrusted code from external repositories can pose a significant security risk, as malicious projects could contain pre/post-build events or malformed data designed to exploit the server's environment. Hardcoded API keys are placeholders, but storing actual keys directly in `appsettings.json` is not recommended for production environments. The system relies on external LLM services, adding a dependency chain for security.
Updated: 2025-12-10GitHub→
0
0
High Cost

Copado-MCP-Server

by 5p875v6f46-ai

Sec4

Automates comprehensive Salesforce process documentation (including validations, business rules, escalations, KPIs, security controls, and visual flow diagrams) using Copado AI's Build Expert model, integrating with LLM clients via MCP (stdio) or HTTP.

Setup Requirements

  • ⚠️Requires active Copado AI API credentials (API Key, Organization ID, Workspace ID), which likely involves a paid service.
  • ⚠️The `COPADO_WORKSPACE_ID` environment variable is required by the `CopadoAPIClient` for proper configuration, but it is not consistently listed in the `.env.example` or documentation setup instructions, leading to potential configuration errors.
  • ⚠️The provided `start-server.sh` script contains hardcoded API credentials; running it as-is will expose these secrets and override `.env` settings.
Review RequiredView Analysis
The `start-server.sh` script, if present and used, hardcodes critical Copado AI API credentials (URL, Key, Org ID, Workspace ID) directly in the file. This is a severe security vulnerability, as these secrets would be exposed if the script is committed to version control or shared. While other parts of the code correctly use environment variables, the existence of this script with hardcoded secrets significantly lowers the overall security assessment. Additionally, the HTTP server uses `cors()` with default wide-open settings, which should be restricted in production environments.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
delian icon

codeguide-mcp

by delian

Sec8

An MCP server that provides coding guides and best practices to AI assistants, designed to extend or replace `AGENTS.md` files.

Setup Requirements

  • ⚠️Requires `config.toml` or environment variables for configuration (e.g., `GUIDES_GITHUB_REPO`, `GUIDES_DIR`).
  • ⚠️Relies on network connectivity to GitHub for fetching remote guides, although local cache and directory fallbacks are supported.
  • ⚠️Python 3.x is required to run the server directly from source.
Verified SafeView Analysis
The server includes basic path traversal prevention for requested guide names (e.g., disallowing '/' and '..'). It performs network requests using `httpx` to the GitHub API, which mitigates generic SSRF risks by hardcoding the domain. Content fetched from GitHub is base64 decoded. There are no clear indications of `eval` or other dynamic code execution patterns. Configuration values, including GitHub repository details, are read from `config.toml` or environment variables, preventing hardcoded secrets. Potential risks include serving malicious Markdown content (client-side rendering vulnerability) or if the GitHub repository configuration could be manipulated to a malicious source within GitHub's ecosystem, but these are outside the direct control of the server's code logic.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
Ali-Hassan2 icon

mcp-ai-agents-lab

by Ali-Hassan2

Sec5

A Model Context Protocol (MCP) server providing tools for AI agents, specifically to add numbers, echo messages, and critically, post content to the X (formerly Twitter) social media platform.

Setup Requirements

  • ⚠️Requires X (Twitter) Developer Account and API Keys (X_API_KEY, X_API_KEY_SECRET, X_ACCESS_TOKEN, X_ACCESS_SECRET) configured as environment variables for the 'x-post-creation' tool to function.
  • ⚠️The server's HTTP endpoint needs proper network security (e.g., firewall, authentication, IP whitelisting) to prevent unauthorized access, especially to the X posting tool.
  • ⚠️The client for this server (`lab-2_(X_Posting_Agent)/mcp-client`) requires a `GEMENI_API_KEY` for LLM integration, incurring external API costs, and a `SERVER_URL` to connect to this MCP server.
Review RequiredView Analysis
The server uses environment variables for sensitive X (Twitter) API credentials (`X_API_KEY`, `X_API_KEY_SECRET`, `X_ACCESS_TOKEN`, `X_ACCESS_SECRET`). However, the `x-tool.js` file explicitly logs these credentials to the console if not found, which is a critical security vulnerability as it can expose secrets in logs. Furthermore, the server uses `StreamableHTTPServerTransport`, exposing its tools over HTTP. Without additional authentication, authorization, or network access controls, this allows any client to call the sensitive X posting tool, posing a significant risk of unauthorized social media activity.
Updated: 2025-12-10GitHub→
0
0
Low Cost
FelipeSchreiber icon

MCP-Server-Oferta

by FelipeSchreiber

Sec8

This server exposes a set of domain-specific tools, resources, and prompts via the FastMCP protocol to be consumed by Large Language Models (LLMs) or other multi-agent automation engines.

Setup Requirements

  • ⚠️Requires Python 3.10+ and dependencies installed via `pip install -r requirements.txt`.
  • ⚠️Requires environment variables in a `.env` file for server settings and optional Azure AD authentication.
  • ⚠️Azure AD authentication (if enabled) requires a configured Azure App Registration (tenant ID, client ID, JWKS URI, issuer, audience).
Verified SafeView Analysis
The server uses `pydantic-settings` for configuration, pulling sensitive information from `.env` files, which is good practice. It supports Azure AD authentication via JWTs, enhancing security. The core server logic (`mcp_server.py` and `services/*.py`) does not contain obvious dangerous patterns like `eval` of untrusted input. However, the provided `Standalone/fastmcp_client_template.py` and `ASGI/fastmcp_client_template.py` (which are client examples, not the server itself) use `eval(json_data)` for parsing SSE responses. While this is client-side code, it's a significant security vulnerability if `json_data` can be controlled by a malicious actor and should be replaced with `json.loads`.
Updated: 2026-01-19GitHub→
0
0
Low Cost
dr-armani icon

mcp-server-demo

by dr-armani

Sec9

This project demonstrates the setup of a FastMCP server, exposing basic tools, resources, and prompt generation functionalities for an agentic system.

Setup Requirements

  • ⚠️Requires Python 3.12 or newer.
  • ⚠️Requires the 'mcp[cli]>=1.22.0' package.
  • ⚠️The recommended run command implies the 'uv' package runner is expected to be installed.
Verified SafeView Analysis
The code uses standard Python f-strings and type hints for basic functions, without direct 'eval' or 'exec'. It does not contain obvious hardcoded secrets. As a network-facing service using 'streamable-http' transport, it has the inherent attack surface of any HTTP server, but the provided application logic is simple and does not introduce specific vulnerabilities.
Updated: 2025-11-30GitHub→
0
0
Low Cost
Yuval-Moag icon

mcp-weather-server

by Yuval-Moag

Sec9

Provides a production-ready Model Context Protocol (MCP) server for integrating AI applications with external weather data via streamable HTTP transport.

Setup Requirements

  • ⚠️Requires a free WeatherAPI.com API key.
  • ⚠️Requires Node.js 20+ for local development.
  • ⚠️Docker and Docker Compose are needed for full-stack deployment.
Verified SafeView Analysis
The project implements strong security practices including environment variable-based secret management (explicitly preventing .env commit), HTTPS enforcement via Traefik, and API timeouts. It also claims non-root container security, though the provided Dockerfile snippet does not explicitly set a non-root user. CORS is set to '*' which is acceptable for a template/PoC but would ideally be tightened in a production environment.
Updated: 2025-11-20GitHub→
0
0
Low Cost
mudinc icon
Sec7

Deploys a remote Model Context Protocol (MCP) server on Cloudflare Workers, providing a set of calculator tools accessible to MCP clients without authentication.

Setup Requirements

  • ⚠️Requires a Cloudflare Workers account for deployment.
  • ⚠️No authentication is required for API access, making it publicly available.
  • ⚠️Requires `npm` and Node.js for local development and deployment.
Verified SafeView Analysis
The server is explicitly designed as 'authless', meaning it intentionally lacks authentication and is publicly accessible. This is a design choice, not a vulnerability, but users must be aware of its open nature. Input validation for the 'add' and 'calculate' tools is performed using Zod, mitigating common injection risks for arithmetic operations. The code does not use 'eval' or other directly dangerous functions based on user input. Environment variables for OAuth and Durable Objects are declared but not directly consumed in the provided source for the core calculator logic, which could be a point of confusion for security auditing if not fully implemented.
Updated: 2025-11-26GitHub→
0
0
Medium Cost
ekfaktldk11 icon

mcp-news-server

by ekfaktldk11

Sec9

An MCP server for fetching US news headlines, summarizing them with AI, and translating them to Korean.

Setup Requirements

  • ⚠️Requires News API Key (Free tier has request limits and provides only recent news)
  • ⚠️Requires OpenAI API Key (Paid service, usage-based billing)
  • ⚠️Node.js and npm are required for installation and execution
Verified SafeView Analysis
API keys are loaded from environment variables, which is good practice. The server makes external API calls to reputable services (News API, OpenAI API) using `axios`. No `eval` or dynamic code execution from user input is present. Inputs to AI models are news text, relying on the LLM's own safety mechanisms. The caching mechanism is simple and does not introduce obvious vulnerabilities. Input sanitization for `get_news_details` relies on the external News API's `q` parameter handling.
Updated: 2026-01-19GitHub→
PreviousPage 668 of 713Next