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)

34
4
Medium Cost

mcp

by plusplusoneplusplus

Sec4

The MCP Server provides an AI-powered framework for command execution, plugin-based tools, secure Python evaluation, and knowledge management, designed for standalone or embedded use.

Setup Requirements

  • ⚠️Requires `uv` for dependency management, which must be installed separately.
  • ⚠️Mandatory `.env` file setup from `config/env.template` with critical environment variables (e.g., `GIT_ROOT`, Azure Repo details).
  • ⚠️External plugins configured in `plugin_config.yaml` require manual `mcp_admin refresh_plugins` execution to be installed.
Review RequiredView Analysis
The server's core functionality involves executing arbitrary commands (`Command Executor`) and Python code (`PyEval`). While `RestrictedPython` is mentioned for `PyEval`, the effectiveness of its security implementation would require thorough auditing. The system also supports loading plugins from arbitrary GitHub repositories via `plugin_config.yaml` and running `mcp_admin refresh_plugins`, posing a significant supply chain risk if not carefully managed. Exposed HTTP/SSE endpoints could also be a vector if not properly secured.
Updated: 2025-11-17GitHub→
34
4
Low Cost
jdlemes icon

MsSqlMCP

by jdlemes

Sec9

Provides an MCP Server for SQL Server database schema inspection and safe, read-only query execution to AI clients.

Setup Requirements

  • ⚠️Requires .NET 10 (or .NET 9 with minor adjustments).
  • ⚠️Requires an accessible SQL Server instance.
  • ⚠️Manual configuration of SQL Server connection string in 'appsettings.json' is mandatory.
Verified SafeView Analysis
The project places a strong emphasis on security by implementing a ReadOnlySqlQueryValidator that blocks 27 common keywords associated with DML, DDL, DCL, and execution commands (e.g., INSERT, UPDATE, DELETE, DROP, EXEC). This validation is backed by 42 dedicated security tests. Connection strings are externalized in 'appsettings.json', preventing hardcoded secrets directly in the codebase. As long as the 'appsettings.json' is securely managed on the host, the application itself is designed to prevent malicious SQL operations.
Updated: 2025-11-28GitHub→
34
4
Medium Cost
mackerelio-labs icon

mcp-server

by mackerelio-labs

Sec9

This server allows AI agents to interact with the Mackerel monitoring and APM platform by exposing its API functionalities as Model Context Protocol tools.

Setup Requirements

  • ⚠️Requires MACKEREL_APIKEY environment variable.
  • ⚠️Requires Node.js and npm (or Docker).
Verified SafeView Analysis
The server correctly handles the Mackerel API key via an environment variable (`MACKEREL_APIKEY`), avoiding hardcoded secrets. There is no usage of 'eval' or any clear obfuscation. Error handling includes a token limit check for responses, which helps prevent excessively large payloads. Standard external API integrations always carry some inherent risks, but best practices are followed.
Updated: 2026-01-14GitHub→
34
2
Medium Cost
erikmaday icon

unusual-whales-mcp

by erikmaday

Sec9

Provides an AI assistant with comprehensive Unusual Whales market data, including options flow, dark pool activity, congressional trades, and more, for natural language market analysis.

Setup Requirements

  • ⚠️Requires an Unusual Whales API Key (available from their website, may be associated with a paid subscription tier).
  • ⚠️Requires Node.js 20.0.0 or higher.
  • ⚠️Requires specific configuration within an MCP-compatible client (e.g., Claude Code/Desktop) to enable the server.
Verified SafeView Analysis
The server uses `process.env.UW_API_KEY` for API authentication, preventing hardcoded secrets. Path parameters are explicitly validated and encoded to prevent path traversal (`/`, `\`, `..`) attempts. It implements robust resilience features like client-side rate limiting and a circuit breaker pattern, enhancing stability and preventing API abuse. While reliance on an external API inherently carries some risk, the server's design focuses on data retrieval rather than arbitrary code execution from API responses, mitigating direct code injection risks.
Updated: 2026-01-19GitHub→
34
2
Medium Cost
SeonghakHong-Sch icon

kStock-mcp-server

by SeonghakHong-Sch

Sec8

This server provides an MCP (Micro-Capability Protocol) interface to interact with the Korean Investment API, enabling functionalities like fetching stock prices, account information, financial data, and executing stock buy/sell orders.

Setup Requirements

  • ⚠️Requires a C++ compiler (e.g., g++ or clang++) and associated build tools.
  • ⚠️Requires external C++ libraries: nlohmann/json, cpp-httplib, and the mcp_server framework (not included in truncated source, assumed external).
  • ⚠️Requires environment variables K_appkey, K_appsecret, K_cano, and K_cano2 to be set for API authentication and account access.
Verified SafeView Analysis
Sensitive information (API keys, account numbers) is loaded from environment variables using `std::getenv`, which is a good practice. Access tokens are obtained via API calls and stored in memory. Request parameters are built from JSON, and `tool::build_query` handles string serialization, reducing direct injection risks. The `mcp::tool_builder` defines parameter schemas, implying input validation by the MCP framework, which is crucial. No direct use of 'eval' or shell command execution from user input is observed. The singleton pattern for `KInvestmentAPI` is noted; while simplifying access, it means credentials are globally shared within the server process, which requires careful management in highly concurrent or multi-tenant scenarios (though not explicitly problematic for the apparent use case).
Updated: 2025-11-26GitHub→
34
3
Medium Cost
Ipiano icon

gdb-mcp

by Ipiano

Sec3

Provides an AI assistant with programmatic access to GDB debugging sessions, enabling debugging, analysis, and control of compiled programs and core dumps.

Setup Requirements

  • ⚠️Requires GDB (GNU Debugger) to be installed and accessible in the system PATH.
  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Requires manual configuration in the MCP client (e.g., Claude Desktop's config file).
  • ⚠️Requires understanding of GDB concepts for effective AI interaction.
Review RequiredView Analysis
This server is inherently high-risk because it exposes highly privileged GDB operations, including arbitrary command execution, function calls within the debugged process, and file system manipulation (via working directory, program/core loading, and GDB commands). The `gdb_execute_command` and `gdb_call_function` tools, while offering command sanitization, allow for direct execution of GDB commands, which can in turn execute shell commands, Python scripts, or access/modify memory and files on the host system. The security of this server relies entirely on the external MCP client implementing robust permissioning, user consent mechanisms, and the server being run in an isolated environment with minimal privileges.
Updated: 2025-12-10GitHub→
34
2
High Cost
serkan-ozal icon
Sec6

AI-powered GitHub analytics platform and chatbot that provides insights into developer profiles, repository statistics, and answers questions using GitHub Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid service)
  • ⚠️Requires GitHub Personal Access Token for GitHub API access
  • ⚠️Python 3.11+ required for the backend
  • ⚠️Node.js >= 18.x and npm/yarn required for the frontend
Verified SafeView Analysis
The backend's FastAPI server is configured with CORS `allow_origins=['*']`, which is a significant security risk in a production environment as it allows requests from any origin. Although API keys (OPENAI_API_KEY, GITHUB_PAT) are properly loaded from environment variables and not hardcoded, the LangChain/LangGraph agent may be vulnerable to prompt injection attacks inherent to LLM-based applications if not specifically mitigated. No `eval` or obvious code obfuscation was detected.
Updated: 2026-01-19GitHub→
34
4
High Cost
rodionlim icon
Sec9

Manages personal financial portfolios including equities, FX, commodities, cash, bonds, and cryptocurrencies, offering AI-driven insights and natural language querying via an MCP server.

Setup Requirements

  • ⚠️Requires Go version 1.24.3 or higher.
  • ⚠️GEMINI_API_KEY is required for AI analytics and MCP functionality (Google Gemini API is a paid service).
  • ⚠️New developers must run `hooks/setup.sh` to install Git pre-commit hooks.
  • ⚠️Google Drive backup requires a one-time OAuth2 Desktop App setup, potentially involving a local loopback server or SSH tunneling for remote deployments.
Verified SafeView Analysis
The project demonstrates a strong security posture regarding secret management, utilizing environment variables for API keys and a `pre-commit` Git hook to prevent accidental commitment of sensitive data. Default server bindings (`0.0.0.0`) for both the main and MCP servers mean they listen on all network interfaces, which requires proper firewall configuration to avoid unintended public exposure. No `eval` or similar code injection vectors are apparent in the Go source code.
Updated: 2026-01-17GitHub→
34
1
Medium Cost
ForceInjection icon

mcp-reference

by ForceInjection

Sec5

A reference processing tool for Markdown documents, providing automated reference extraction, citation renumbering, bibliography generation, and related literature recommendation functions, integrated via Model Context Protocol (MCP) into Trae IDE.

Setup Requirements

  • ⚠️Requires Python 3.12
  • ⚠️Requires `uv` package manager to be installed
Review RequiredView Analysis
The server exposes endpoints (`/api/v1/extract/references`, `/api/v1/extract/citations`, etc.) that accept a `file_path` parameter. This path is directly used by `app/services/markdown_parser.py`'s `read_markdown` function via `open()`, which makes the server vulnerable to arbitrary file read (path traversal) if not properly isolated or if input paths are not sanitized. The `/api/v1/system/stop` endpoint allows unauthenticated shutdown, which can disrupt service, though this is less critical than the file read vulnerability. It also uses `os._exit(0)` for an ungraceful shutdown.
Updated: 2025-11-24GitHub→
34
1
Medium Cost
ruslanmv icon

webclone

by ruslanmv

Sec8

Clones and archives entire websites, including those requiring authentication, accessible via command line, a desktop GUI, or as an AI agent tool through Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires Chrome/Chromium browser to be installed for Selenium functionality.
  • ⚠️Requires Python 3.11+ environment.
  • ⚠️Requires `mcp` library for AI agent (MCP) functionality (`make install-mcp`).
Verified SafeView Analysis
The server's core function involves interacting with external websites using a real browser (Selenium) and making HTTP requests. This carries inherent risks if malicious URLs are provided by the user/AI agent, as the browser can execute arbitrary JavaScript. However, the implementation includes input validation (Pydantic models) and explicit warnings about cookie file security. Sensitive session cookies are stored locally and are explicitly git-ignored. The `save_authentication` tool intentionally offloads manual browser interaction to the GUI/CLI to prevent blocking AI agents. Subprocess calls are limited to opening local file explorers. The code itself does not contain 'eval', obfuscation, or hardcoded secrets.
Updated: 2025-11-24GitHub→
34
3
Medium Cost
schlpbch icon

aareguru-mcp

by schlpbch

Sec9

Provides real-time and historical Swiss Aare river data, including water temperature, flow rates, and safety assessments, to AI assistants via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires Python 3.10+.
  • ⚠️The Aareguru API is for non-commercial use only and requires attribution.
  • ⚠️Usage of 'uv' for dependency management is recommended, but Docker is also supported for setup.
Verified SafeView Analysis
The server implements rate limiting for HTTP endpoints and includes configurable API key authentication (though not enabled by default). It does not appear to use dangerous patterns like 'eval' or allow arbitrary code execution. Standard HTTP server security practices should be applied in deployment, such as using HTTPS and securing the host environment.
Updated: 2026-01-18GitHub→
34
29
High Cost
linw1995 icon

nvim-mcp

by linw1995

Sec3

A Model Context Protocol (MCP) server that provides seamless integration with Neovim instances, enabling AI assistants to interact with the editor through connections and access diagnostic information.

Setup Requirements

  • ⚠️Requires Neovim (version 0.11.3+ is mentioned for integration testing) to be installed and accessible in the system's PATH.
  • ⚠️For 'Dynamic Tool System' and optimal auto-connection, the `nvim-mcp` Lua plugin must be installed within Neovim's configuration (e.g., via `lazy.nvim`).
  • ⚠️Many LSP-related tools (e.g., `lsp_code_actions`, `lsp_hover`) require specific Language Servers (e.g., `gopls` for Go, `ts_ls` for TypeScript) to be configured and running within the connected Neovim instance.
Review RequiredView Analysis
The server includes an `exec_lua` tool that allows executing arbitrary Lua code within the connected Neovim instance. While this is an intended feature for powerful integrations, it poses a critical security risk. If the server is exposed via HTTP (`--http-port` with `--http-host 0.0.0.0`) without robust external authentication/authorization, a malicious actor gaining network access to the server could execute arbitrary code within the user's Neovim, potentially compromising the editor and underlying system. There are no explicit hardcoded secrets found in the provided code.
Updated: 2026-01-02GitHub→
PreviousPage 155 of 760Next