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
dchu3 icon
Sec9

The MCP server wraps the Base Blockscout API to provide agent-friendly tools for querying account, transaction, contract, token, log, and DEX router activity on Base Mainnet and Base Sepolia.

Setup Requirements

  • ⚠️Requires Node.js 20+ to run.
  • ⚠️Requires manual creation and configuration of a '.env' file based on '.env.example'.
  • ⚠️Blockscout API access is necessary, and heavy usage might require an API key to avoid rate limits.
Verified SafeView Analysis
The server appears well-designed for security, using environment variables for API keys and configurable values (like router addresses), adhering to read-only operations, and implementing rate limiting, caching, and retries for external API interactions. There is no usage of 'eval' or other inherently dangerous patterns. Router addresses are hardcoded as defaults but are explicitly overrideable via a JSON file and CLI, mitigating potential single-point-of-failure concerns.
Updated: 2025-11-22GitHub→
0
0
Medium Cost
workademy icon
Sec3

Provides comprehensive integration with the Workademy LMS platform, enabling LLMs to manage courses, users, enrollments, learning paths, and generate reports.

Setup Requirements

  • ⚠️Requires Node.js 18 or higher to run.
  • ⚠️Requires a Workademy account with API access and an API key.
  • ⚠️Authentication involves a hardcoded system credential, posing a significant security risk.
Review RequiredView Analysis
CRITICAL: The server's authentication mechanism for obtaining an access token uses a hardcoded Basic Authentication credential ('Basic c3lzdGVtOkViNFRRaXpEdFlpYndJTHI=' which decodes to 'system:Eb4TQizDtYibwILr'). This static, hardcoded system credential is a severe security vulnerability, as it is exposed in the source code and could be exploited to compromise the token issuance process or gain unauthorized access if not properly secured on the Workademy API side. Additionally, the server interacts with the local file system (using `fs/promises` and `path`) to save and analyze CSV reports in a Claude-specific temporary directory, which introduces a potential vector for file system manipulation if not handled robustly.
Updated: 2025-12-10GitHub→
0
0
Medium Cost
ryan-rbw icon
Sec9

Provides GitLab integration tools to AI assistants like Claude Desktop and ChatGPT, acting as a secure bridge for code and project management operations.

Setup Requirements

  • ⚠️Requires Python 3.12+ or Docker for deployment.
  • ⚠️Requires access to a GitLab instance and the ability to register an OAuth application (admin rights recommended). This involves several manual steps to configure Client ID, Secret, Redirect URI, and Scopes in GitLab.
  • ⚠️Persistent user sessions with encrypted token storage require generating and securely managing a Fernet encryption key (`KEPLER_MCP_TOKEN_ENCRYPTION_KEY`).
Verified SafeView Analysis
The server implements strong security practices including OAuth 2.0 with PKCE for user authentication, Fernet encryption for session tokens at rest, and explicit redaction of sensitive data in logs. It uses Pydantic's SecretStr for secret configuration fields. The server binds to '0.0.0.0' by default, which is noted as intentional for Docker container networking, but administrators must ensure it's behind a reverse proxy or firewall in production to mitigate direct exposure. There are no obvious malicious patterns, 'eval' usage (other than standard library imports, not arbitrary execution) or hardcoded secrets in the provided code.
Updated: 2025-12-08GitHub→
0
0
Low Cost
chienchuanw icon

gma2-mcp

by chienchuanw

Sec3

Provides a programmatic interface via Telnet to control grandMA2 lighting consoles using the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires grandMA2 lighting console with Telnet access enabled.
  • ⚠️Requires Python 3.12+.
  • ⚠️Login credentials (GMA_HOST, GMA_USER, GMA_PASSWORD) must be configured in a .env file or environment variables; otherwise, insecure defaults are used.
Review RequiredView Analysis
The server relies on Telnet, an unencrypted protocol, meaning all communication including login credentials and commands are transmitted in plain text, making it vulnerable to eavesdropping and credential theft. Default login credentials ('administrator', 'admin') are hardcoded, which, if not overridden in the `.env` file, pose a significant security risk. Unauthorized access could lead to disruption of lighting control systems.
Updated: 2025-12-26GitHub→
0
0
Medium Cost
keyurgolani icon

TasksMcp

by keyurgolani

Sec8

Enterprise-grade task management and orchestration for AI agents, featuring dependency management and multi-interface support.

Setup Requirements

  • ⚠️Requires Node.js runtime environment.
  • ⚠️Local development requires 'npm install' and 'npm run build'.
  • ⚠️Relies heavily on environment variables for configuration, necessitating careful setup (e.g., data storage paths, database credentials).
  • ⚠️Default data storage is filesystem-based, requiring a configurable DATA_DIRECTORY.
Verified SafeView Analysis
The server uses robust Zod for input validation and comprehensive error handling with retry logic and circuit breakers. Input preprocessing is applied for agent-friendly type coercion before strict validation. File system operations are wrapped in file locks and retry logic. No direct use of 'eval' or obvious command injection vulnerabilities were found. Configuration is managed via environment variables and dedicated config files, avoiding hardcoded secrets. Rate limiting is configured but the middleware implementation wasn't fully shown in the provided snippets. Overall, it follows good security practices for its context.
Updated: 2026-01-19GitHub→
0
0
High Cost
a7medsa22 icon

Ai-Knowledge-Hub

by a7medsa22

Sec8

AI-powered knowledge management and research platform featuring document management, intelligent summarization, Q&A, and MCP integration.

Setup Requirements

  • ⚠️Requires Docker and Docker Compose for PostgreSQL.
  • ⚠️Requires local Ollama server running and a model pulled (e.g., 'llama3.1:8b') for local AI, or an OpenAI/Anthropic API key for cloud AI.
  • ⚠️Requires Node.js 18+ and npm.
Verified SafeView Analysis
The application uses NestJS with security best practices like Helmet for HTTP headers, global validation pipes, JWT-based authentication with ownership checks for resources, and throttling for rate limiting. Sensitive configurations (API keys, JWT secrets, DB URLs) are correctly loaded from environment variables. A custom HTTP exception filter provides consistent error handling. File uploads use unique filenames, and the `serveFile` endpoint relies on Express's `res.sendFile` which typically prevents path traversal. However, production deployments should harden CORS policies beyond `localhost:3000` and confirm robust path traversal prevention in `serveFile` for filenames provided by users. No 'eval' or obvious obfuscation detected.
Updated: 2026-01-12GitHub→
0
0
Medium Cost
trilogy-group icon

brainlift-mcp-server

by trilogy-group

Sec8

An agent-facing server that provides tools to manage and query a user's BrainLift knowledge base, primarily by interacting with the BrainLift API via Google OAuth.

Setup Requirements

  • ⚠️Requires Python 3.14 or newer.
  • ⚠️Requires a Google Cloud Project OAuth 2.0 Client ID setup for authentication. The `setup.sh` script automates this if specific environment variables (`API_KEY`, `API_BASE_URL`, `HIVE_INSTANCE_ID`) are provided, otherwise, manual OAuth setup and initial authentication flow are required.
  • ⚠️The `BRAINLIFT_API_URL` environment variable must be set.
Verified SafeView Analysis
The server uses standard Python libraries for HTTP requests (`requests`) and Google OAuth (`google-auth-oauthlib`). It handles OAuth credentials by storing them in local files (`client-secrets.json`, `.gcp-saved-token.json`), with paths configurable via environment variables. The `setup.sh` script retrieves and formats OAuth credentials from an external API (`$API_BASE_URL`) using `curl` and `jq`. The main security considerations are the trustworthiness of the `API_BASE_URL` and `API_KEY` provided to the `setup.sh` script, as well as the secure storage of the generated OAuth credential files in the deployment environment. No 'eval', code obfuscation, or obvious malicious patterns were found. Input validation for tool parameters (`brainlift_id`, `dok_levels`) is present. Network requests have timeouts.
Updated: 2025-12-03GitHub→
0
0
High Cost
darthvader3010 icon
Sec2

An image classification system using EfficientNet-B7 for categorizing images and providing predictions via a Flask API and an MCP server.

Setup Requirements

  • ⚠️Requires Node.js 14+ for the MCP server.
  • ⚠️EfficientNet-B7 model weights (~250MB) are downloaded on the first run, leading to a longer initial startup.
  • ⚠️GPU is highly recommended for faster inference; CPU performance can be 1-2 seconds per image.
  • ⚠️Python 3.8+ is required.
Review RequiredView Analysis
CRITICAL VULNERABILITY: The /classify endpoint in `app.py` directly uses user-provided `image_path` from `request.json` to open files (`Image.open(image_path)`). This is a severe Local File Inclusion (LFI) vulnerability, allowing an attacker to attempt to read arbitrary files on the server's filesystem. There is also no sanitization for uploaded filenames, though `os.path.join` offers some protection. The Flask server is exposed on `0.0.0.0`, which is risky if not behind a firewall.
Updated: 2025-11-30GitHub→
0
0
Low Cost
mtrmarko icon

cat-facts-mcp

by mtrmarko

Sec3

Provides an API endpoint for retrieving random cat facts.

Review RequiredView Analysis
Only README.md was provided as source code. The actual server implementation code is missing, making a comprehensive security audit impossible. Cannot assess for 'eval', network risks, hardcoded secrets, or malicious patterns within the server's logic.
Updated: 2025-12-02GitHub→
0
0
Low Cost
arunshenoy99 icon

mcp-http-bridge

by arunshenoy99

Sec9

Bridges STDIO-based Model Context Protocol (MCP) clients to HTTP(S)-based MCP server endpoints, enabling integration with tools like Cursor and Claude Desktop.

Setup Requirements

  • ⚠️Requires Node.js >= 18.0.0
  • ⚠️Requires MCP_ENDPOINT environment variable to be set
Verified SafeView Analysis
The project explicitly states 'Zero dependencies - uses only Node.js built-ins', which significantly reduces supply chain attack surface. Sensitive configuration (custom headers) is managed via environment variables. No 'eval' or obfuscation found. As a bridge, its security is primarily dependent on the upstream MCP endpoint it connects to.
Updated: 2025-11-27GitHub→
0
0
Low Cost
yuu-git icon
Sec9

Provides integrated GitHub/local file operations, Git management, and knowledge base access (via Notion/Docusaurus) to multiple IDEs and AI clients through a C#/.NET Model Context Protocol (MCP) server.

Setup Requirements

  • ⚠️Requires .NET 8.0 SDK, Git, and a GitHub Personal Access Token (PAT) for full GitHub/Git functionality.
  • ⚠️Claude Desktop integration requires absolute paths in its configuration, unlike VS Code/Visual Studio which support relative paths.
  • ⚠️Visual Studio 2026 Preview requires manual approval for third-party MCP servers due to enhanced security.
Verified SafeView Analysis
The server implements good security practices, such as recommending `appsettings.local.json` for secrets and explicitly discouraging hardcoding. It also explicitly disables `AutoPull/AutoPush` by default to prevent unwanted Git history pollution, requiring explicit Git commands from the user/AI. Visual Studio 2026 Preview integration includes an additional security layer requiring manual approval for third-party MCP servers. The primary risks are inherent to giving an AI programmatic access to file systems and Git operations if misconfigured or misused by the AI itself.
Updated: 2025-12-04GitHub→
0
0
Low Cost
UnitVectorY-Labs icon

mcp-shopping-list-firestore

by UnitVectorY-Labs

Sec9

A lightweight, Firebase-backed MCP server for management of a grocery list via simple CRUD operations.

Setup Requirements

  • ⚠️Requires a Google Cloud Project ID.
  • ⚠️Requires a Firestore Database name.
  • ⚠️Requires Google Cloud authentication credentials (e.g., via `gcloud` CLI, environment variables, or a credentials JSON file specified with the `--credentials` flag) to access Firestore.
Verified SafeView Analysis
The server demonstrates good security practices for its scope: it avoids hardcoded secrets by relying on environment variables, uses context with timeouts for external calls, and performs basic input validation. There are no 'eval' or similar dangerous patterns detected. As with any network service, external deployment security (firewalls, access control) would be necessary.
Updated: 2026-01-17GitHub→
PreviousPage 478 of 713Next