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
Medium Cost

mcp-kit

by omar391

Sec6

This project appears to be a TypeScript/JavaScript development kit or server application, leveraging modern build and testing tools (Rsbuild, Vitest), likely for a multiplayer or backend system.

Setup Requirements

  • ⚠️Requires a Node.js environment (likely a recent LTS version) and a package manager (npm/yarn/pnpm) to install dependencies.
  • ⚠️A build step via Rsbuild (e.g., 'npm run build') is necessary to compile TypeScript and bundle assets before running the server.
  • ⚠️Likely requires configuration via environment variables or specific configuration files (e.g., .env) for server port, database connections, or other operational parameters.
Review RequiredView Analysis
Without code access, a thorough security audit for 'eval' or obfuscation is impossible. As a server, it inherently exposes network services, making it susceptible to standard web/server vulnerabilities if not properly secured. The presence of package.json suggests third-party dependencies, which could also introduce supply chain risks.
Updated: 2025-11-18GitHub→
0
0
Low Cost
jck411 icon
Sec8

A Model Context Protocol (MCP) server for controlling Spotify playback, searching, queue management, and library access.

Setup Requirements

  • ⚠️Requires a Spotify Developer account and creation of a Spotify App to obtain Client ID and Client Secret.
  • ⚠️A one-time local OAuth flow must be completed using `scripts/get_spotify_token.py` to acquire a Spotify Refresh Token.
  • ⚠️For Cloud Run deployment, specific Google Cloud APIs (Cloud Run, Secret Manager, Cloud Build) must be enabled, and IAM permissions granted for the Cloud Run service account to access secrets.
Verified SafeView Analysis
The `scripts/get_spotify_token.py` utility contains hardcoded `CLIENT_ID` and `CLIENT_SECRET` values. While the primary server implementation correctly uses environment variables and Google Secret Manager for credentials, this script's hardcoded values could be a source of confusion or a potential security oversight if not replaced with user-specific credentials during the OAuth flow. The server itself manages refresh tokens and retrieves client credentials from secure environment variables. A `TODO` exists to persist new refresh tokens to a database/Secret Manager, which is a future enhancement but not a current security vulnerability.
Updated: 2025-11-27GitHub→
0
0
Low Cost
oryce icon

mai-ml-5-sem

by oryce

Sec9

Building and training a decision tree model for binary classification tasks, handling both real and categorical features.

Setup Requirements

  • ⚠️Requires Python 3.x environment
  • ⚠️Requires `numpy` library (`pip install numpy`)
Verified SafeView Analysis
The code primarily implements a decision tree algorithm using standard Python libraries (numpy, collections). There are no indicators of network operations, file system access beyond typical module loading, use of `eval` or `exec`, hardcoded credentials, or other patterns commonly associated with severe security vulnerabilities. The `feature_types` input is validated. The risk is primarily related to the input data provided to the model, which is outside the scope of this code's direct control.
Updated: 2025-12-14GitHub→
0
0
High Cost
mkmlab-hq icon

athena-mcp-servers

by mkmlab-hq

Sec7

A collection of essential AI microservices designed to power intelligent agents and automate development tasks through a multi-agent control protocol.

Setup Requirements

  • ⚠️Requires Python 3.11+.
  • ⚠️Requires `pip install -r requirements.txt` which includes many AI/DB dependencies (e.g., mcp SDK, psycopg2, qdrant-client, sentence-transformers, google-generativeai).
  • ⚠️Requires API keys for Google Gemini (GEMINI_API_KEY or GOOGLE_API_KEY) and potentially other third-party services.
  • ⚠️Requires active PostgreSQL and Qdrant vector databases (local or remote/VPS).
  • ⚠️Relies on external 'SPICE Hybrid Brain' infrastructure hosted by MKM Lab, accessed via a hardcoded IP address.
Verified SafeView Analysis
The project generally follows good security practices by using environment variables for API keys and sensitive configurations (e.g., POSTGRES_URL, GEMINI_API_KEY, QDRANT_URLs). However, some external service endpoints (e.g., GEMMA_VPS_URL, VPS_QDRANT_URL) are hardcoded with IP addresses, which introduces a reliance on specific external infrastructure and requires trusting the operators of those IPs. The default 'postgres:password' for POSTGRES_URL in example code is a common weak credential, which users must be careful to override in their .env for production environments. No 'eval' or obvious malicious patterns were found.
Updated: 2025-11-29GitHub→
0
0
Low Cost
aiscibe-mcp-deployment icon

mcp-cf7b72a7-10c103b7-todo-list-mcp

by aiscibe-mcp-deployment

Sec8

Acts as a Model Context Protocol (MCP) server to provide a Todo List functionality by integrating with an external API.

Setup Requirements

  • ⚠️Requires the `mcp` Python library (e.g., `pip install -r requirements.txt`).
  • ⚠️No functional tools are implemented in the provided source code, requiring manual implementation for actual use by an MCP client.
  • ⚠️The `call_api` helper contains a bug where `conn_name='default'` is used, but 'default' is not defined in `API_CONNECTIONS`, making API calls non-functional as-is.
Verified SafeView Analysis
No hardcoded secrets or obvious malicious patterns found. The `call_api` function, while not directly exposed to user input in the provided code, could enable SSRF if tools are later implemented without proper input validation, allowing arbitrary API calls to `jsonplaceholder.typicode.com`. The `conn_name='default'` bug in `call_api` (which is not in `API_CONNECTIONS`) prevents it from being functional as-is, inadvertently reducing its attack surface. Error logging includes stack traces, which could reveal internal paths in production.
Updated: 2025-11-29GitHub→
0
0
High Cost
Gitmy3 icon
Sec5

Develop a Python-based server that integrates Wazuh SIEM alerts with an LLM (preferably OpenAI) for enhanced security analysis and response.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid)
  • ⚠️Requires a running Wazuh SIEM instance for integration
  • ⚠️Requires Python environment setup
Review RequiredView Analysis
A comprehensive security audit cannot be performed as only the `README.md` file was provided. The project involves integration with an LLM and a SIEM, which typically entails network communication and API key handling. Without source code, it's impossible to check for 'eval', obfuscation, hardcoded secrets, or malicious patterns. Score is neutral due to lack of information.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
koonwen icon

dev-mcp-server

by koonwen

Sec8

Provides OCaml Language Server Protocol (LSP) and ocp-index tools for AI assistants to analyze and interact with OCaml codebases.

Setup Requirements

  • ⚠️Requires `ocamllsp` binary to be installed and available in PATH.
  • ⚠️Requires `ocp-index` and `ocp-grep` binaries for full functionality (ocp-index tool group).
  • ⚠️Python 3.10+ and `uv` package manager must be installed for installation and execution.
Verified SafeView Analysis
The server executes external OCaml tools (ocamllsp, ocp-index, ocp-grep) via subprocesses. While arguments are passed as lists to mitigate shell injection, the security relies on the trustworthiness and robustness of these external binaries against crafted inputs. No 'eval' or hardcoded secrets found within the Python code.
Updated: 2025-12-29GitHub→
0
0
High Cost
felix-toledo icon
Sec6

An orchestration system for AI agents to interact with custom tools via the Model Context Protocol (MCP), integrating Large Language Models (LLMs) like OpenAI with backend services.

Setup Requirements

  • ⚠️Requires Node.js v20.x or higher and npm v10.x or higher.
  • ⚠️Requires PostgreSQL v14 or higher, which needs to be deployed externally as it is not included in the Docker Compose setup.
  • ⚠️Requires an OpenAI API Key, which typically involves a paid subscription to OpenAI services.
  • ⚠️Docker Desktop/Engine is required for the recommended Docker Compose setup.
Verified SafeView Analysis
The MCP Server (`services/mcp-server`) has `cors({ origin: '*' })` enabled by default, which is an overly permissive configuration and a significant security risk for production environments, as it allows any domain to make requests. The Orchestrator uses `JSON.parse(toolCall.arguments)` on LLM-generated strings; while `JSON.parse` itself is generally safe, robust schema validation (using Zod) within each custom tool is critical to prevent malformed inputs from causing unexpected application behavior or errors. Authentication to the Orchestrator is enforced via a `KAA` header, relying on a shared secret from environment variables, which is a good practice. Sensitive API keys and database credentials are correctly sourced from environment variables. The system's prompt for the LLM is verbose, but prompt injection remains an inherent risk in LLM-based systems.
Updated: 2025-12-03GitHub→
0
0
Low Cost
LetsVenture2021 icon

dealengine-mcp-nano

by LetsVenture2021

Sec6

Provides a secure and abstracted interface for an application (likely 'DealEngine') to interact with the Notion API, enabling programatic data management within Notion databases, pages, and blocks.

Setup Requirements

  • ⚠️Requires Notion API Key (obtain from Notion Developer Console)
  • ⚠️Requires Node.js runtime
  • ⚠️Requires `.env` file for configuration
Verified SafeView Analysis
The server uses environment variables for sensitive API keys (NOTION_API_KEY), which is good practice. It handles errors and logs details. However, there is no explicit input validation or sanitization of the `req.body` payloads before they are passed directly to the Notion API. This means a malicious or malformed request from an upstream client could potentially lead to unexpected behavior, API errors, or rate limit issues on the Notion side. The `console.error` logging of `err.response?.data` could also expose internal Notion API error structures if not managed in a production environment.
Updated: 2026-01-01GitHub→
0
0
Low Cost
themeselection icon

notes-mcp-demo

by themeselection

Sec9

A Model Context Protocol server for managing notes, offering tools for CRUD operations, a markdown snapshot, and an LLM summarization prompt.

Setup Requirements

  • ⚠️Requires Node.js (LTS) and npm installed.
  • ⚠️Understanding of Model Context Protocol (MCP) clients is necessary to fully utilize the server's tools and prompts.
  • ⚠️Notes data is stored locally in `~/.notes-mcp/data/notes.json`.
Verified SafeView Analysis
The server operates locally via standard I/O (stdio), significantly reducing network attack surface. Input validation is performed using Zod schemas for all tool inputs. Data is stored in local files under the user's home directory (`~/.notes-mcp/data/notes.json`). No sensitive hardcoded credentials, use of 'eval', or other obviously malicious patterns were detected. The local file storage could be a concern if highly sensitive data is stored, but it's appropriate for a 'notes' application.
Updated: 2025-11-24GitHub→
0
0
Low Cost
julienld icon
Sec8

The MCP Server Tester is a FastMCP-based harness designed to launch, proxy, and manage other MCP servers over stdio for testing and development purposes.

Setup Requirements

  • ⚠️Requires an MCP client (e.g., Codex CLI) to interact with the server's control tools.
  • ⚠️Child servers managed by this tester must be MCP-compliant and communicate over stdio.
Verified SafeView Analysis
The server's core function involves launching arbitrary commands and passing environment variables to subprocesses based on user input. While `shlex.split` is used for command parsing, direct execution of user-provided commands (e.g., `tester_control_start_server`) means that if the harness itself is exposed to untrusted input, it presents a command injection risk. However, this is an inherent feature for a testing harness designed to control child servers, not a vulnerability in its intended use by trusted developers in a controlled environment. No explicit 'eval' or hardcoded secrets were found in the `mcp-server-tester`'s source code. The demo credentials in the install script are for a separate child server.
Updated: 2025-12-05GitHub→
0
0
High Cost
tsoernes icon
Sec8

Provides a robust Model Context Protocol (MCP) server for executing shell commands, managing background jobs, detecting developer tools, and ensuring security within a development environment.

Setup Requirements

  • ⚠️Requires Rust 1.70+ and Cargo to build from source.
  • ⚠️Optional 'Sudoers Timestamp Sharing' involves modifying `/etc/sudoers.d/` which requires root privileges and understanding of sudo security implications (e.g., `!tty_tickets` configuration).
  • ⚠️Default askpass script `~/scripts/askpass-zenity.sh` implies that `zenity` (or a similar GUI tool) must be installed and available for interactive sudo priming if sudoers sharing is not configured.
Verified SafeView Analysis
The server implements a strong default denylist for dangerous commands (`rm -rf /`, `shutdown`, fork bombs, etc.), supports PTY for isolation, and allows custom denylist patterns. Sudo handling with automatic `-n` (non-interactive) wrapping and a keepalive mechanism is a key feature for convenience, but the `!tty_tickets` sudoers configuration (if applied as recommended) explicitly notes a security trade-off by allowing any process running as your user to reuse your sudo timestamp. This requires careful consideration and a reasonable `timestamp_timeout` for maximum security. No obvious hardcoded secrets or 'eval' vulnerabilities were found. Overall, the server is designed with security in mind, but the sudo integration requires responsible configuration.
Updated: 2025-12-31GitHub→
PreviousPage 361 of 760Next