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
Low Cost
mistermusiker icon

DLVMCP

by mistermusiker

Sec9

The server sets up basic TCP and WebSocket listeners to simulate a Minecraft server, demonstrating network communication handling for game-like applications.

Setup Requirements

  • ⚠️Requires Node.js installed.
  • ⚠️Requires 'ws' npm package to be installed (npm install ws).
  • ⚠️Ports 25565 (TCP) and 8080 (WebSocket) must be open and available.
Verified SafeView Analysis
The current implementation is very basic, primarily setting up network listeners without sophisticated data parsing or state management. This simplicity inherently limits immediate security risks. There are no 'eval' calls, obfuscation, or hardcoded secrets. Future development to implement actual Minecraft protocol handling would introduce new attack surface areas requiring careful validation and sanitization.
Updated: 2025-12-14GitHub→
0
0
Medium Cost
maldalur icon
Sec6

Automate grading of Moodle submissions using AI, track student progress, and generate reports.

Setup Requirements

  • ⚠️Requires Ollama to be installed and running locally or at a specified host for AI model inference.
  • ⚠️Requires significant Moodle Web Service configuration (creating a service, assigning capabilities, and generating a token).
  • ⚠️Requires `MOODLE_URL`, `TOKEN_MOODLE`, `COURSE_LIST`, `OLLAMA_HOST`, and `OLLAMA_MODEL` environment variables to be configured in a `.env` file.
Review RequiredView Analysis
The `moodle_client.py`'s `download_file` and `_process_vpl_file_entry` functions are vulnerable to path traversal. If a malicious filename containing '..' (e.g., `../../etc/passwd`) is received from the Moodle API, it could lead to arbitrary file writes outside the intended 'downloads' directory. Additionally, while environment variables are used for sensitive API tokens, the `MOODLE_SERVICE_SETUP.md` document explicitly shows a token value, which is a common security risk if copied directly into production code. AI model output is parsed from JSON, which is generally safe, but if displayed on an unsanitized web interface, could lead to XSS (not directly applicable to this project's current output methods to Markdown files).
Updated: 2025-11-27GitHub→
0
0
Medium Cost
immy2good icon
Sec3

Provides MQL4/MQL5 development tools, including compilation, syntax validation, code analysis, AI/ML data collection, and deep documentation integration, for professional MetaTrader development workflows.

Setup Requirements

  • ⚠️Requires MetaTrader 4 or 5 to be installed on the host system (or Wine configured for Linux/macOS).
  • ⚠️Requires Node.js >=16.0.0.
  • ⚠️The conflicting documentation regarding 'interface provider only' vs. 'direct tool execution' is a critical gotcha for deployment and security expectations.
Review RequiredView Analysis
The core server (`mql4-tools-server.js`) directly executes external processes (e.g., `metaeditor.exe`, Python scripts) and performs extensive file system operations (read/write). This contradicts explicit documentation and configuration claiming the server acts as an 'interface provider only' with 'no process execution'. User-provided `filePath` parameters in various tools (`compile_mql4`, `validate_syntax`, `analyze_code`, `auto_fix_syntax_errors`, etc.) could potentially be exploited for arbitrary file access (read/write) or, in the case of `compile_mql4`, arbitrary code execution via `metaeditor.exe` if not properly sandboxed. This discrepancy between code and documentation makes it unsafe for deployments expecting an interface-only security model.
Updated: 2025-11-28GitHub→
0
0
Medium Cost
olgasafonova icon

gleif-mcp-server

by olgasafonova

Sec9

Enables AI assistants to verify legal entities, research corporate ownership structures, and perform compliance checks by querying the Global Legal Entity Identifier (LEI) database.

Setup Requirements

  • ⚠️Requires Go 1.25+ to build from source.
  • ⚠️macOS requires manual permission adjustments (`chmod +x`, `xattr -d com.apple.quarantine`) for downloaded binaries to run.
Verified SafeView Analysis
The server primarily acts as a proxy to GLEIF's public API, which requires no authentication. No hardcoded secrets were identified in the provided source code. It employs good practices for external API interaction including client-side rate limiting, retry logic with exponential backoff, and robust error handling. Requests use `http.NewRequestWithContext` and `slog` for structured logging. The build script uses `CGO_ENABLED=0`, enhancing binary security. Overall, the implementation adheres to good security principles for its intended function.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
ahmetatar icon

mcp2browser

by ahmetatar

Sec9

Enables AI assistants to programmatically interact with web pages through a Model Context Protocol (MCP) server for browser automation tasks.

Setup Requirements

  • ⚠️Requires Playwright browser binaries to be downloaded (typically handled by `npm install`, but can sometimes require specific system dependencies or manual intervention if auto-download fails).
Verified SafeView Analysis
The server uses Playwright to launch a headless browser and navigate to arbitrary URLs via the 'openPage' tool. While this is its intended functionality, exposing this server to an untrusted network would allow remote execution of browser actions, posing a significant security risk. For its intended local use with AI assistants, the code appears robust with basic URL validation for the 'openPage' input.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
arthurfantaci icon

zoom_transcript_mcp

by arthurfantaci

Sec8

Manages and provides an interface for interacting with Zoom Cloud Recording transcripts, allowing listing, downloading, searching, and action item extraction.

Setup Requirements

  • ⚠️Requires a Zoom Account with Cloud Recording enabled.
  • ⚠️Requires creation of a Zoom Server-to-Server OAuth App in the Zoom App Marketplace with specific `cloud_recording:read` scopes to obtain Account ID, Client ID, and Client Secret.
  • ⚠️Node.js v16 or higher is required.
Verified SafeView Analysis
The server correctly uses environment variables for sensitive Zoom API credentials. It interacts with the Zoom API over HTTPS and performs file system operations (read/write transcripts and metadata) to a designated directory, with path sanitization in `formatFileName`. No `eval` or arbitrary command execution is detected. The security primarily relies on the Zoom OAuth configuration and the underlying security of the Zoom API itself. The default `TRANSCRIPTS_DIR` to `process.cwd()/transcripts` could be a minor concern if the working directory is not secure, but it can be configured.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
krakend icon

mcp-server

by krakend

Sec9

Universal MCP server providing intelligent assistance for KrakenD API Gateway configuration validation, security auditing, and configuration generation.

Setup Requirements

  • ⚠️Optimal functionality (e.g., precise version-aware validation and auditing) often requires the `krakend` binary or `docker` to be installed and accessible in your system's PATH.
  • ⚠️Depending on the installation method and target directory (e.g., `~/.local/bin`), you may need to manually add the server's installation path to your system's PATH environment variable.
Verified SafeView Analysis
The server runs locally via stdio, exposing no network ports and performing no remote access beyond downloading documentation and JSON schemas from `krakend.io`. File system access is limited to the current working directory for configuration files (read-only) and a dedicated data directory (`~/.krakend-mcp/` or `data/`) for cache and search index (read/write). No hardcoded secrets or 'eval' equivalent found in the analyzed source code. It operates with user permissions, not elevated privileges. The `SECURITY.md` explicitly outlines supported versions, vulnerability reporting, and user best practices, contributing to a strong security posture.
Updated: 2026-01-09GitHub→
0
0
High Cost

BoostHub

by Anderson-SG

Sec9

An MCP (Model Context Protocol) server designed to boost developer productivity by offering AI-driven code review prompts, code rule generation, and utility tools for integration with clients like Cursor.

Setup Requirements

  • ⚠️Requires Docker for running pre-configured MCP servers (e.g., sequentialthinking, playwright) when using the `/create-mcp-config` prompt, or for standard Docker deployment.
  • ⚠️Requires `MCP_API_KEY` environment variable for secure operation in 'production' environments.
  • ⚠️Requires Node.js and Bun runtime (or Docker) if running locally from source.
Verified SafeView Analysis
The server implements API key authentication via `Authorization: Bearer` or `X-API-Key` headers. The `MCP_API_KEY` environment variable is explicitly enforced as mandatory in 'production' environments, which is a strong security practice. In 'development' mode, authentication is optional, which could be a minor risk if accidentally deployed without a key in production. No 'eval' or direct obfuscation found. Sensitive information like API keys are managed via environment variables. Proper logging for unauthorized access attempts is in place.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
glennart8 icon

MCP_mail_server

by glennart8

Sec8

Automates email handling for a fictional timber company by using AI to classify emails and execute predefined tools for responses and actions.

Setup Requirements

  • ⚠️Requires a Gemini API Key for AI classification and response generation (paid service).
  • ⚠️Requires manual Google OAuth 2.0 configuration (creating a project, enabling Gmail API, generating credentials.json) to interact with real Gmail.
  • ⚠️Requires Python 3.x and installing dependencies from requirements.txt.
Verified SafeView Analysis
The system uses `pickle` for persisting Google OAuth tokens (`token.pickle`). While used for legitimate credential storage, deserializing data from `pickle` can pose a security risk if the file is maliciously tampered with or loaded from an untrusted source. However, in this context, the file is self-generated by the application for its own trusted credentials.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
rameshcreations icon

docker-mcp

by rameshcreations

Sec6

This project implements a Model Context Protocol (MCP) server that exposes Docker management capabilities as MCP tools, allowing connection to clients like Amazon Q or Claude Desktop for AI-driven container orchestration.

Setup Requirements

  • ⚠️Requires Docker Engine to be installed and running on the host system.
  • ⚠️Requires Python 3.10+.
  • ⚠️The client configuration paths (e.g., `mcp/venv/bin/python`, `mcp/docker-mcp.py`) assume the repository is cloned into a `~/mcp` directory; users must adjust these paths if cloned elsewhere.
  • ⚠️Requires an MCP-compatible client (e.g., Amazon Q, Claude Desktop) to interact with the server.
Verified SafeView Analysis
The server itself does not contain obvious malicious patterns, hardcoded secrets, or use dangerous functions like 'eval'. However, it exposes powerful Docker management capabilities (e.g., `run_container`, `delete_container` with `force`) directly to an MCP client. If the MCP client or the host running this server is compromised, it could lead to arbitrary code execution or full control over the Docker host. The security risk lies in the inherent power of the exposed tools and the permissions with which the Docker daemon operates, rather than specific vulnerabilities in the server's code.
Updated: 2025-12-02GitHub→
0
0
Low Cost
hyfoxus icon

movieRecsBot

by hyfoxus

Sec8

The service exposes an IMDb vector database through Model Context Protocol (MCP) endpoints, allowing clients to perform natural language movie searches and fetch movie metadata.

Setup Requirements

  • ⚠️Requires a local Ollama instance running to provide embeddings.
  • ⚠️Requires a PostgreSQL database with the pgvector extension and pre-loaded IMDb data.
  • ⚠️Default `SPRING_DATASOURCE_PASSWORD` and `APP_ADMIN_BOOTSTRAP_TOKEN` are 'changeme'/'bootstrap-token' respectively, requiring explicit modification for secure deployment.
  • ⚠️Initial setup involves running `scripts/setup_data.sh` to download and embed IMDb data, which can be time-consuming and resource-intensive.
Verified SafeView Analysis
The server uses named parameters for most SQL queries, which is a good practice against SQL injection. Sensitive administrative endpoints are protected by an `X-Bootstrap-Token`. Docker secrets are utilized for the Telegram bot token (though not directly for this MCP server, it's a monorepo setup). However, default placeholder values like `change-me` for `APP_ADMIN_BOOTSTRAP_TOKEN` and `SPRING_DATASOURCE_PASSWORD` are present; these must be explicitly changed in production environments. There is a minor risk with direct string concatenation for `hnsw.ef_search` and `maxTitles` in SQL queries, although these values are currently derived from application configuration properties, not directly from untrusted user input.
Updated: 2026-01-17GitHub→
0
0
Low Cost
devonfox icon

notes-mcp

by devonfox

Sec8

An MCP server for creating markdown notes and managing tasks, sending them to Discord and optionally saving locally.

Setup Requirements

  • ⚠️Requires Node.js 18+ for native FormData and Blob support.
  • ⚠️A Discord Webhook URL is required for core functionality, which needs to be created and configured.
  • ⚠️Environment variables (e.g., `DISCORD_WEBHOOK_URL`, `LOCAL_FILE_PATH`, `TASKS_FILE_PATH`) must be correctly set up.
Verified SafeView Analysis
The server correctly uses environment variables for sensitive information like `DISCORD_WEBHOOK_URL`, preventing hardcoded secrets. User-provided `title` for markdown notes is sanitized to prevent path traversal issues when forming filenames. File paths for local saving (`LOCAL_FILE_PATH`, `TASKS_FILE_PATH`) are configured via environment variables, which is standard practice. No direct `eval` or shell command execution from user input is observed. Network requests to Discord include timeouts. Overall, the code follows good security practices for its described functionality, assuming secure configuration of environment variables. The primary risks would stem from misconfiguration of local file paths or improper management of the Discord webhook URL.
Updated: 2025-11-18GitHub→
PreviousPage 598 of 713Next