Stop Searching. Start Trusting.

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

Browse Directory Why 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)

Medium Cost

mcp-debugserver

by nickzer0

Sec7

Provides a robust interface for an LLM or agent to debug Windows processes using WinDbg/CDB.

Setup Requirements

⚠️Requires Python 3.10+.
⚠️Requires Windows Debugging Tools (cdb.exe, dbgsrv.exe) to be installed from the Windows SDK.
⚠️Requires administrator privileges to attach to many processes.
⚠️The 'arch' environment variable (x86/x64) must be set to match the target process architecture to avoid RPC errors.

Verified SafeView Analysis

The server acts as a wrapper around WinDbg/CDB, which inherently requires high system privileges (e.g., administrator rights) to attach to and debug processes. This makes the tool powerful and, if compromised or misused by an untrusted agent, capable of significant system impact. The codebase itself does not contain obvious malicious patterns like 'eval' or hardcoded credentials. It relies on 'subprocess.Popen' for execution of external debuggers and 'os.kill' for sending break signals, which are legitimate operations for a debugging tool. The primary security consideration is the inherent power of debugging tools and the necessary elevated permissions required for their operation.

Updated: 2025-12-01GitHub→

Low Cost

rpg-mcp-server

by optimalhorror

Sec9

Manages D&D-style RPG campaign data, including combat, NPCs, and bestiary entries, for AI assistance.

Setup Requirements

⚠️Requires Python 3.12 or newer.
⚠️Requires 'uv' (a modern Python package manager) for installation and running.
⚠️Utilizes local JSON files for all data storage by default, not a database, requiring proper file system management.
⚠️Designed to be run as an MCP server, typically connected to an AI agent like Claude Desktop, rather than a standalone application.

Verified SafeView Analysis

The server primarily handles local file I/O for data persistence within a dedicated 'campaigns' directory. File paths are safely constructed using a `slugify` function, which effectively prevents directory traversal attacks. There are no 'eval', 'exec', or direct network exposure points (communication is via standard I/O using MCP). No hardcoded secrets were found. The primary security considerations would be file system permissions in the deployment environment and ensuring the integrity of the local JSON data files.

Updated: 2025-11-26GitHub→

Low Cost

mcp-gitlab-mcp-server-pibe

by bioanywhere

Sec9

Deploys and manages a generic MCP server application using Docker.

Setup Requirements

⚠️Docker required
⚠️Port 3000 must be available on host

Verified SafeView Analysis

The provided Docker deployment configuration (`docker-compose.yml`, `run.sh`) adheres to strong security practices: no hardcoded secrets, 'eval' or malicious patterns observed. Secrets for CI/CD (`GCP_CREDENTIALS`, `GCP_PROJECT_ID`) are explicitly managed outside the repository. The container runs on a standard non-privileged port and uses a volume for data persistence. A full audit would require the application's Dockerfile and source code, but the deployment setup itself is secure.

Updated: 2025-12-02GitHub→

Low Cost

weather-MCP-server

by MadanReddy6

Sec9

This server provides a tool to fetch current weather information for a specified location using the OpenWeatherMap API.

Setup Requirements

⚠️Requires OpenWeatherMap API Key (free tier available)
⚠️Requires Python 3.x
⚠️Requires 'httpx', 'python-dotenv', and 'fastmcp' Python libraries

Verified SafeView Analysis

The API key (OPENWEATHER_API_KEY) is properly loaded from environment variables via python-dotenv, preventing hardcoding of secrets. No 'eval' or similarly dangerous functions are used. Network requests are made using the standard 'httpx' library. The application appears secure for its intended purpose.

Updated: 2025-12-02GitHub→

Low Cost

repo-mcp-server

by malconip

Sec8

An AI-powered code intelligence system that stores, searches, and analyzes structured knowledge about code files to integrate with Claude Desktop.

Setup Requirements

⚠️Requires a Supabase account for PostgreSQL database (free tier available), including manual steps to create the project and retrieve the connection string.
⚠️A `MCP_SECRET_KEY` must be generated (e.g., using `openssl rand -hex 32`) and explicitly set as an environment variable.
⚠️Requires Python 3.11+ to run.

Verified SafeView Analysis

The system correctly uses environment variables for sensitive data like `DATABASE_URL` and `MCP_SECRET_KEY`, with deployment instructions emphasizing encrypted secrets. SQLAlchemy is used for database interactions, providing protection against common SQL injection vulnerabilities. CORS is implemented via `ALLOWED_ORIGINS` to restrict access. A warning is issued if the default `MCP_SECRET_KEY` is in use. No 'eval' or direct arbitrary command execution from user input is apparent.

Updated: 2025-12-01GitHub→

Medium Cost

revolution-mcp-server

by zweischritte

Sec7

Serves as a knowledge context protocol (MCP) server, managing and exposing a specialized knowledge corpus for AI models.

Setup Requirements

⚠️Requires Node.js and npm

Verified SafeView Analysis

Exposes a knowledge corpus via stdio and Streamable HTTP on port 3000, requiring REVOLUTION_HTTP_ENABLE=true for HTTP access. Standard network security practices should be applied if exposed externally.

Updated: 2025-11-23GitHub→

High Cost

api-mapping

by MadsRunge

Sec4

This project automates the reverse-engineering and documentation of private website APIs using an AI agent with Chrome DevTools.

Setup Requirements

⚠️Requires Google Chrome or Chromium installed and accessible for Chrome DevTools control.
⚠️Requires an AI model endpoint, likely needing a Gemini API Key (Paid) or a locally running Gemini-compatible model.

Review RequiredView Analysis

The project employs an AI agent with Chrome DevTools for browser control, enabling extensive network requests to third-party websites for reverse-engineering their private APIs. This inherently carries risks such as potential violations of terms of service, IP blocking, rate limiting, and unpredictable behavior from the AI agent's interactions with dynamic web content. There is no code provided to inspect for 'eval' or obfuscation within the agent's logic, but the nature of external interaction introduces inherent security considerations.

Updated: 2025-11-17GitHub→

Medium Cost

playlist-matcher

by gauravdewani99

Sec8

Automatically organizes liked Spotify songs into existing playlists based on artist and genre similarity, and can run on a schedule.

Setup Requirements

⚠️Requires creating a Spotify Developer app and obtaining a Client ID from the Spotify Developer Dashboard.
⚠️Manual configuration of `~/.claude.json` with the server's command path and the `SPOTIFY_CLIENT_ID` environment variable is needed.
⚠️Initial authentication requires an interactive browser-based OAuth flow to grant Spotify access.
⚠️The scheduled job setup (using `launchd`) is specific to macOS and requires manual command execution for installation and management.

Verified SafeView Analysis

The server uses secure OAuth PKCE for Spotify authentication. Tokens are stored locally in the user's config directory, which is a standard practice for desktop applications. The `run-cron.sh` script, while hardcoding a `SPOTIFY_CLIENT_ID`, does not expose a critical secret as Spotify Client IDs for public clients (using PKCE) are not considered confidential. No 'eval' or malicious patterns were identified.

Updated: 2026-01-18GitHub→

Medium Cost

sleeper-mcp

by draffetseder

Sec9

An MCP tool server exposing the Sleeper fantasy sports API, enabling AI models and agents to manage and query fantasy leagues.

Setup Requirements

⚠️Requires Node.js v20.0.0 or higher.

Verified SafeView Analysis

The server uses axios with a fixed base URL to the public Sleeper API, which significantly mitigates Server-Side Request Forgery (SSRF) risks. Input arguments are directly interpolated into the URL path for Sleeper API endpoints; while the Sleeper API itself is expected to handle path sanitization, a highly malformed input could potentially access unexpected public Sleeper API endpoints within the same domain. There is no usage of 'eval', 'child_process', or direct file system access, and no hardcoded secrets were found. Data returned from the Sleeper API is stringified and relayed, so its content depends on what the public Sleeper API exposes.

Updated: 2026-01-19GitHub→

Low Cost

MCP-servers-with-Typescript

by shraddhamahajan93

Sec7

Implements a basic Minecraft Protocol (MCP) server in TypeScript, allowing clients to connect and interact through defined protocol phases such as handshake, status, login, and play.

Setup Requirements

⚠️Requires Node.js runtime to execute
⚠️Requires a TypeScript development environment (ts-node, tsc) for direct execution or compilation

Verified SafeView Analysis

The server is configured to listen on '0.0.0.0' (all network interfaces) by default, making it publicly accessible. The critical 'handleData' packet parsing logic is not fully visible and represents a potential area for vulnerabilities if not robustly implemented against malformed input. No 'eval' calls, obfuscation, or hardcoded sensitive secrets were found in the provided code snippets.

Updated: 2026-01-19GitHub→

Medium Cost

trace-mcp

by Mnehmos

Sec7

Static analysis engine for detecting schema mismatches between data producers (MCP tools) and consumers (client code), offering code generation and continuous validation.

Setup Requirements

⚠️File System Access & Permissions: The tool requires read/write access to specified project and source code directories. Care must be taken with `init_project` and `comment_contract` tools as they modify files and create directories, and path parameters should be trusted or carefully sanitized to prevent unintended file modifications outside the intended project scope.
⚠️Language Support Discrepancy: While the `init_project` tool's schema suggests support for 'go' and 'rust' languages, the provided source code only includes parsers for 'typescript', 'python' (with basic regex parsing), and 'json_schema'. Users attempting to configure for Go or Rust will find these features unimplemented.
⚠️Computational Intensity: For very large codebases or during active development in `watch` mode, the static analysis can be CPU and memory intensive, potentially impacting development machine performance. Caching helps but re-analysis of changed files is still required.

Verified SafeView Analysis

The server performs extensive local file system operations (read/write) based on user-provided paths for `rootDir`, `filePath`, `producerDir`, `consumerDir`, and `projectDir`. While `path.resolve` is used for normalization, an attacker with local access and appropriate file system permissions could exploit path traversal vulnerabilities (e.g., using `../../`) to read or modify files outside the intended project directories, especially with `addContractComments` (which modifies source files) and `init_project` (which creates project structures). The `json_schema` parser reads arbitrary JSON files, which could lead to denial-of-service if extremely large or malformed JSON is supplied. Communication is via standard I/O (stdio), which limits direct network exposure, but careful handling of user input is critical if integrated into other services.

Updated: 2025-12-11GitHub→

Low Cost

MCP-Manager

by muzahirabbas

Sec9

Manages Model Context Protocol (MCP) server configurations across multiple local JSON files via a desktop GUI.

Setup Requirements

⚠️Requires Python 3.8+
⚠️Requires PySide6 library (`pip install PySide6`)

Verified SafeView Analysis

The application primarily interacts with local files chosen by the user. It uses standard JSON parsing (`json.load`, `json.loads`) and file system operations (`os.makedirs`, `shutil.copy2`, `os.replace`). There are no obvious remote execution vulnerabilities, network risks, hardcoded secrets, or malicious patterns like `eval()`.

Updated: 2026-01-17GitHub→

PreviousPage 633 of 760Next