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
Low Cost
tyukei icon
Sec9

Provides IT study group and event information from Connpass API as an MCP server.

Setup Requirements

  • ⚠️Requires Connpass API Key (free, but registration with Connpass is required)
  • ⚠️Requires Python 3.12 or higher
  • ⚠️Setup instructions use `uv` as the package manager
Verified SafeView Analysis
The project adheres to good security practices by loading the Connpass API key from environment variables (CONNPASS_API) rather than hardcoding it. It uses the `httpx` library for secure HTTP requests to a known API endpoint (connpass.com). There are no visible uses of `eval`, `exec`, or direct shell commands with unsanitized user input. The internal rate limiting further mitigates potential abuse or accidental overloading of the external API.
Updated: 2025-12-14GitHub→
0
0
High Cost
puran-water icon

site-fit-mcp-server

by puran-water

Sec2

MCP server for constraint-based site layout optimization of wastewater/biogas facilities, generating diverse layouts while respecting engineering rules and process topology.

Setup Requirements

  • ⚠️Requires Python 3.11+.
  • ⚠️Optional features like GIS file loading, DXF export, and PDF generation require additional `pip install` commands (e.g., `pip install 'site-fit-mcp[gis]'` for GIS support).
  • ⚠️When running in HTTP mode with viewer (`site-fit-serve`), it requires `uvicorn` and `fastapi`.
Review RequiredView Analysis
CRITICAL: The `sitefit_load_gis_file` tool directly passes a user-controlled `file_path` to `fiona.open` without sanitization. This allows for arbitrary file read vulnerabilities (e.g., path traversal like `../../../etc/passwd`). CRITICAL: The `sitefit_export_pack` tool accepts a user-controlled `output_dir` for writing generated files (e.g., DXF, PDF, CSV). This allows for arbitrary file write vulnerabilities (e.g., path traversal to write to sensitive system directories). LOW RISK: The `rules_override` parameter in `sitefit_generate` is passed to `yaml.safe_load` for merging into the RuleSet. While `safe_load` mitigates many YAML deserialization attacks, it's not entirely immune, especially with older `PyYAML` versions or complex gadget chains. User-controlled input to a YAML parser is generally discouraged if the parser is not fully sandboxed.
Updated: 2026-01-10GitHub→
0
0
High Cost
blueorbitz icon

jedi-mcp

by blueorbitz

Sec8

Converts technical documentation websites into Model Context Protocol (MCP) servers, enabling AI coding assistants to access detailed framework, library, API, SDK, and language documentation during development.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Requires a Google Gemini API key or AWS Bedrock credentials (paid AI services are used extensively).
  • ⚠️Browser-based navigation extraction (required for some sites like Microsoft Learn or those with JavaScript-rendered navigation) depends on Playwright, which needs its browsers installed after `pip install playwright`.
Verified SafeView Analysis
The project demonstrates good security practices for web crawling (rate limiting, URL validation, external link filtering) and AI API key management (environment variables, no hardcoded secrets, parameterized SQL queries). The use of Playwright for headless browser automation (to handle JavaScript-rendered content) inherently carries a risk of executing JavaScript from crawled sites; however, this is mitigated by robust URL filtering and strict adherence to documentation-specific content.
Updated: 2025-12-25GitHub→
0
0
Low Cost
pawneetdev icon

dataforseo-mcp

by pawneetdev

Sec9

This server acts as a Model Context Protocol (MCP) adapter, exposing DataForSEO's Research Toolkit API to Claude Desktop as automatically generated tools for SEO research tasks.

Setup Requirements

  • ⚠️Requires DataForSEO API username and password to be configured in a .env file.
  • ⚠️The 'generate_registry.py' script must be run first to create the necessary tool registry and endpoint files before starting the server.
  • ⚠️Requires Python 3.13 or newer.
Verified SafeView Analysis
Credentials are loaded from environment variables (.env file) and not hardcoded, which is good practice. The server interacts with external APIs (DataForSEO, GitHub for OpenAPI spec), which is standard for its function. No 'eval' or obvious malicious patterns were found in the provided code snippets. Logging is directed to stderr to avoid interfering with stdout JSON-RPC communication.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
AshGhimire icon

mcp-chess-server

by AshGhimire

Sec9

Provides access to Chess.com player data for profiles and statistics through an MCP server.

Setup Requirements

  • ⚠️Requires Python 3.14 or higher (this is a future Python version and not yet released, meaning it cannot be run on current stable Python versions like 3.8-3.12).
  • ⚠️Requires the `uv` package manager (or `uvx` executable) to be installed for running via the provided MCP configuration.
Verified SafeView Analysis
The server acts as a wrapper for the public Chess.com API. It uses the `requests` library with `raise_for_status()` for error handling. No hardcoded secrets, `eval()` calls, or obvious malicious patterns were found. The 'User-Agent' string is slightly malformed ('moizillla' instead of 'mozilla'), which is a minor stylistic issue but not a security vulnerability.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
djdjdj08 icon
Sec9

An MCP server providing an AI agent with tools to ingest content, search a knowledge base, and log completion results for an educational context.

Setup Requirements

  • ⚠️Requires Node.js 20.18.0
  • ⚠️The 'ingest_content', 'search_content', and 'log_completion_result' tools are currently stubs and require a separate backend implementation for full functionality.
Verified SafeView Analysis
The server uses Express.js and Zod for schema validation, which are good practices. Error handling for MCP requests is present. There are no obvious signs of 'eval' usage, hardcoded secrets, or malicious patterns. The `TODO` comments indicate that actual backend integrations (for ingest, search, log) are pending, and their security would depend on their implementation. General web security practices (e.g., HTTPS, rate limiting) would be needed for production deployment, but the provided code itself is robust.
Updated: 2025-12-02GitHub→
0
0
High Cost
nathanstitt icon

shotter

by nathanstitt

Sec2

Automating iOS Simulator interactions and screenshot capture for AI assistants via the Model Context Protocol.

Setup Requirements

  • ⚠️Requires macOS with Xcode installed (for iOS Simulator)
  • ⚠️Requires Node.js 18+
  • ⚠️Requires Facebook IDB (`brew install idb-companion`)
  • ⚠️Requires ANTHROPIC_API_KEY (paid API for AI agent functionality)
Review RequiredView Analysis
The server allows execution of arbitrary shell commands via the 'runBefore' field in workflow YAML files (src/workflow/runner.ts). The 'runBefore' script is spawned with `shell: /bin/bash` and directly uses the provided string from the workflow configuration. This is a critical remote code execution vulnerability if a malicious or untrusted workflow file is ever processed. Additionally, `xcrun simctl` commands are executed using `execSync` (src/utils/simulator.ts), though input for these is less directly user-controlled.
Updated: 2025-12-15GitHub→
0
0
High Cost
iarwain1 icon
Sec2

Enables AI coding agents to interact with the ErgoAI reasoning engine for logic programming, query execution, and knowledge representation.

Setup Requirements

  • ⚠️Requires ErgoAI (a separate, complex knowledge representation and reasoning system) to be pre-installed and configured on the system.
  • ⚠️May require manually setting the `ERGOAI_PATH` environment variable if ErgoAI is not installed in a standard location.
Review RequiredView Analysis
Critical shell injection vulnerability identified: The `run_ergo_query` tool directly uses user-provided `query` strings in a `spawn` command where `shell: true` is enabled for Windows (via `spawn` options). This allows malicious shell metacharacters (e.g., `&`, `|`, `&&`, `||`) within the query to execute arbitrary commands via `cmd.exe` on Windows systems. On Linux/macOS, direct command injection into `runergo` itself (if it supports such escapes within its '-e' argument) is also a concern. While `run_ergo_code` and `check_ergo_syntax` write code to temporary files, `run_ergo_file` directly uses `file_path` as an argument, which could still be vulnerable to shell injection depending on the path's content. The overall lack of input sanitization for user-provided query strings and file paths before executing external processes makes this server unsafe to run without modifications.
Updated: 2025-12-10GitHub→
0
0
Low Cost
Nik95 icon
Sec9

This server implements a Spring AI Micro-Agent Protocol (MCP) tool server, demonstrating a custom Jackson ObjectMapper configuration for serialization/deserialization.

Setup Requirements

  • ⚠️Requires Java Development Kit (JDK) 17+.
  • ⚠️Requires Maven or Gradle to build and run.
Verified SafeView Analysis
The code is simple and straightforward, defining a basic MCP tool with custom Jackson ObjectMapper configuration. No 'eval' or dynamic code execution is present. Input is handled as simple strings, mitigating common injection risks. No hardcoded secrets or suspicious network activity are visible. 'failOnUnknownProperties(false)' is used, which could be a deserialization risk in more complex scenarios with untrusted inputs and vulnerable gadgets, but for this specific, simple tool, it does not pose an immediate or critical threat.
Updated: 2025-11-20GitHub→
0
0
Medium Cost

cas-jobs-sdss-mcp

by sciserver

Sec8

A foundational template for building M.C.P. servers, specifically geared towards scientific data processing, likely related to SDSS (Sloan Digital Sky Survey) within the SSEC-JHU ecosystem.

Setup Requirements

  • ⚠️Requires Docker for containerized deployment.
  • ⚠️Optimized for Python 3.11.
  • ⚠️Requires running 'python project_setup.py' for initial template configuration.
Verified SafeView Analysis
No 'eval' or obfuscation found in the provided description. The template integrates security practices like Bandit scans and GitHub Security workflows. It defines network exposure (e.g., port 8000) for the server, which is an expected part of its operation.
Updated: 2025-11-17GitHub→
0
0
Low Cost

n8n-mcp-server

by janitor

Sec8

Deploys a remote Model Context Protocol (MCP) server on Cloudflare Workers for exposing AI tools without requiring authentication.

Setup Requirements

  • ⚠️Requires a Cloudflare account for deployment.
  • ⚠️The server lacks authentication, making it unsuitable for production use with sensitive tools or data.
  • ⚠️Connecting local clients like Claude Desktop requires installing and running the 'mcp-remote' proxy.
Verified SafeView Analysis
Runs on Cloudflare Workers, which provides a secure, sandboxed execution environment. However, the server is explicitly designed to operate 'Without Auth,' meaning any client with its URL can access and execute its defined tools. This makes it suitable for demos, personal projects, or internal tools where access control is not critical, but highly unsuitable for production environments involving sensitive data or operations. No obvious use of 'eval' or obfuscation based on the file list and README.
Updated: 2025-11-17GitHub→
0
0
Medium Cost
NoSugarCoffee icon

youmind-mcp

by NoSugarCoffee

Sec8

Provides a Model Context Protocol (MCP) server to retrieve content from YouMind by craft ID for use with AI models.

Setup Requirements

  • ⚠️Requires a YouMind account and manual extraction of the 'YOUMIND_AUTH_TOKEN' cookie from the browser.
  • ⚠️Designed to be used with an existing MCP client (e.g., Claude Desktop, Cursor).
  • ⚠️Requires Python 3.8 or higher.
Verified SafeView Analysis
The server uses environment variables for authentication tokens, which is good practice. It handles common HTTP errors gracefully. No malicious patterns like 'eval' or hardcoded secrets were found. The primary security consideration for users is the sensitive nature of the 'YOUMIND_AUTH_TOKEN', which is a session cookie and requires careful handling to prevent unauthorized access to their YouMind account.
Updated: 2026-01-17GitHub→
PreviousPage 558 of 760Next