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
PurpleSquirrelMedia icon

ibmz-mcp-server

by PurpleSquirrelMedia

Sec4

Integrate Claude Code with IBM Z mainframe capabilities for HSM-backed key management and exposing z/OS applications as REST APIs, enabling AI-powered enterprise operations and security.

Setup Requirements

  • ⚠️Requires an IBM Cloud account with active Key Protect service (and potentially other IBM services like NLU, watsonx.ai, Cloudant, TTS for full demo functionality).
  • ⚠️Requires access to an IBM mainframe with z/OS Connect Enterprise Edition (EE) installed and configured for using z/OS Connect tools.
  • ⚠️Crucial environment variables (IBM_CLOUD_API_KEY, KEY_PROTECT_INSTANCE_ID, ZOS_CONNECT_URL, ZOS_CONNECT_USERNAME, ZOS_CONNECT_PASSWORD) must be configured correctly; the hardcoded secrets in demo files should NOT be used in a production or shared environment.
Verified SafeView Analysis
The repository contains numerous hardcoded API keys and instance IDs within its demo scripts, which is a significant security risk as it can lead to accidental exposure or misuse if not properly managed. While the main server (`index.js`) correctly uses environment variables for sensitive configuration, the presence of hardcoded secrets in related files is a major concern. The `zos_connect_call_service` tool allows constructing arbitrary requests to a configurable mainframe URL; although path parameters are encoded and payloads are JSON stringified, this tool presents a potential Server-Side Request Forgery (SSRF) risk if `ZOS_CONNECT_URL` can point to internal networks, and a risk of exploiting vulnerabilities in target mainframe applications if Claude is prompted to craft malicious API calls.
Updated: 2025-12-15GitHub→
0
0
High Cost
Tsarri icon
Sec8

A multi-agent Retrieval-Augmented Generation (RAG) system for legal tech, focusing on document intelligence, deadline extraction, and strategic analytics.

Setup Requirements

  • ⚠️Requires Supabase account with PostgreSQL and pgvector extension.
  • ⚠️Requires Anthropic API Key (Paid API usage for Claude models).
  • ⚠️Requires Google Gemini API Key (Paid API usage for Gemini models; optional but disables preprocessing/validation if missing).
  • ⚠️Requires Claude Desktop (for MCP server integration).
Verified SafeView Analysis
The server handles API keys via environment variables, which is good practice. File system operations for document storage and deletion are restricted to client-specific subdirectories, mitigating arbitrary file access risks. The Supabase client library is used, which typically prevents SQL injection. However, as an LLM-powered application, it is inherently susceptible to prompt injection attacks against the underlying Claude and Gemini models, where a malicious input could potentially manipulate AI behavior. No direct use of `eval` or obvious obfuscation was found.
Updated: 2025-12-23GitHub→
0
0
Low Cost
Sec5

Provides server-side infrastructure for Unity-based multiplayer games, likely managing game state, player connections, and other backend services.

Setup Requirements

  • ⚠️Requires Unity Editor for development, compilation, or initial project setup due to Unity project files detected.
  • ⚠️Requires appropriate network configuration (e.g., port forwarding, firewall rules) for external access, typical for any server application.
  • ⚠️Specific backend dependencies (e.g., database, external services) are unknown without a README or code analysis.
Review RequiredView Analysis
A detailed code-level security audit is not possible without access to actual code files. Server applications inherently carry network risks (e.g., DDoS, unauthorized access). The absence of a README means no explicit security practices or known vulnerabilities are documented, requiring caution when deploying.
Updated: 2025-11-17GitHub→
0
0
Low Cost
linogarciavallejo icon

octopus-mcp-poc

by linogarciavallejo

Sec9

A proof-of-concept server designed to receive and log webhooks from GitHub and GitLab repositories.

Setup Requirements

  • ⚠️Requires Node.js and npm/yarn to install dependencies and run.
  • ⚠️Requires environment variables (PORT, WEBHOOK_SECRET, GITLAB_WEBHOOK_SECRET, APP_ID) to be set, typically via a .env file.
  • ⚠️For webhooks to reach it, the server must be publicly accessible or run behind a tunneling service (e.g., ngrok).
Verified SafeView Analysis
The server uses established webhook middleware libraries (`@octokit/webhooks` and `gitlab-webhook-middleware`) which are designed to verify webhook signatures using provided secrets, mitigating common spoofing risks. The current code primarily logs payloads and does not contain `eval`, obfuscation, or direct use of user-controlled input in shell commands. Hardcoded secrets are avoided through the use of `dotenv`. The main security consideration would be if further, unprovided logic were to act upon the logged webhook payloads in an insecure manner.
Updated: 2025-11-19GitHub→
0
0
Medium Cost
BridgerB icon

metabase-mcp

by BridgerB

Sec8

Provides an MCP server to interact with Metabase instances, enabling programmatic access to databases, tables, saved questions (cards), dashboards, and collections, as well as raw API calls.

Setup Requirements

  • ⚠️Requires Deno runtime to execute.
  • ⚠️Requires a running Metabase instance reachable by the server.
  • ⚠️Requires either METABASE_API_KEY or (METABASE_USERNAME and METABASE_PASSWORD) environment variables for Metabase authentication, as used by `src/client.ts` (note: README uses METABASE_BASE_URL, but client.ts expects METABASE_URL).
Verified SafeView Analysis
The server correctly retrieves Metabase API credentials (API key or username/password) from environment variables, avoiding hardcoded secrets. It implements a session token management with expiration and retry. The 'api' tool allows arbitrary Metabase API calls, which is powerful but also broadens the potential attack surface if the agent using it is compromised; however, this is a feature, not a vulnerability in the server's implementation. No direct use of 'eval' or other highly dangerous functions are observed in the provided source code, with queries being executed against the Metabase API rather than locally.
Updated: 2025-12-03GitHub→
0
0
Low Cost
Sec6

This server acts as a bridge, allowing external applications to control Adobe Photoshop through its ExtendScript engine, facilitating automation and custom workflows.

Setup Requirements

  • ⚠️Requires Node.js environment installed.
  • ⚠️Requires Adobe Photoshop installed and running.
  • ⚠️Requires configuration based on `mcp-config-example.json` (e.g., creating `mcp-config.json`).
Verified SafeView Analysis
Without code inspection, cannot verify input validation or presence of 'eval'. As a server interacting with Photoshop via ExtendScript, there's a potential risk of arbitrary script execution if inputs are not properly sanitized before being passed to the ExtendScript engine. It should primarily be used in a trusted, local environment.
Updated: 2025-11-18GitHub→
0
0
Medium Cost
cardmagic icon

slack-messages

by cardmagic

Sec9

Fuzzy search and browse Slack messages and threads from the command line or as an MCP server.

Setup Requirements

  • ⚠️Requires Node.js 22+
  • ⚠️Requires a Slack User OAuth Token (xoxp-) with multiple specific scopes (`channels:history`, `groups:history`, `im:history`, `mpim:history`, `users:read`, `channels:read`, `groups:read`, `im:read`, `mpim:read`). Creating this token involves creating a Slack app and manually adding scopes.
  • ⚠️Initial indexing (`slack-messages index`) can take a long time for large Slack workspaces due to Slack API rate limits and the volume of messages.
Verified SafeView Analysis
The server stores Slack user OAuth tokens locally in `~/.slack-messages/config.json`. This is necessary for its functionality, and the README explicitly warns users not to share or commit these tokens. Sensitive input (Slack token) is handled securely via a password prompt. Database operations use parameterized queries, mitigating SQL injection risks. Slack API calls are rate-limited and use the official SDK, which handles retries. No 'eval', obfuscation, hardcoded secrets, or overtly malicious patterns were found in the provided source code.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
lascam-UFU icon

swechain-mcp-server

by lascam-UFU

Sec8

Orchestrates interactions with a SweChain blockchain, providing a Model Context Protocol (MCP) interface to manage auctions, bids, balances, and query blockchain status for programmatic access.

Setup Requirements

  • ⚠️Requires a running SweChain node (from `lascam-UFU/swechain` repository)
  • ⚠️Requires the `swechaind` binary to be installed and available in the system's PATH
  • ⚠️Uses `--keyring-backend test` for all `swechaind` interactions, implying a test/development keyring setup.
Verified SafeView Analysis
The server primarily operates by executing the `swechaind` command-line utility with arguments derived from user input. While `exec.CommandContext` is used (which mitigates direct shell injection by passing arguments separately), the security critically depends on `swechaind`'s robust handling of its arguments. Input validation (e.g., `isValidCosmosAddress`, numeric checks) is present, reducing common risks. The `--keyring-backend test` flag suggests a development/testing environment, which is less secure for production. No 'eval' or direct malicious patterns were found. Overall, the code appears safe assuming the underlying `swechaind` binary handles its parameters securely.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
luisfelipesena icon

webots-youbot-mcp

by luisfelipesena

Sec7

Provides a Model Context Protocol (MCP) bridge to allow AI agents like Claude Code and Cursor to monitor and control Webots robot simulations in real-time.

Setup Requirements

  • ⚠️Requires Python 3.10+.
  • ⚠️Requires 'mcp', 'pydantic', and optionally 'Pillow' Python packages.
  • ⚠️On macOS, force reload/reset functionality requires granting accessibility permissions to the terminal/IDE running the server for 'osascript' to control Webots.
  • ⚠️Requires a running Webots simulation with the 'mcp_bridge.py' integrated into the robot controller.
Verified SafeView Analysis
The server uses file-based inter-process communication (via status.json and commands.json) within a designated 'data' directory. This mechanism, while central to MCP, means local file system access could be a vector for malicious commands if compromised. A notable concern is the use of 'subprocess.run' to execute 'osascript' commands on macOS for 'force reload' and 'force reset' functionalities. This grants the script the ability to send keyboard shortcuts directly to the Webots application, requiring the user to grant accessibility permissions to the terminal or IDE running the server. No 'eval' or direct network listeners are present, and no hardcoded secrets were found.
Updated: 2025-12-25GitHub→
0
0
Medium Cost
dawkareshubham icon

mcp-server-study

by dawkareshubham

Sec3

A collection of Model Context Protocol (MCP) projects demonstrating various server implementations and client interactions, including tools, resources, prompts, multi-server connections, and OpenAI integration.

Setup Requirements

  • ⚠️Requires `uv` for dependency management and running projects.
  • ⚠️Requires Python 3.11+ (some projects specify 3.10+, but 3.11+ is common).
  • ⚠️OpenAI API Key is required for projects integrating with LLMs (e.g., mcp-client, mcp-build-memory-tracker, mcp-build-client-agent-airbnb-memory).
Review RequiredView Analysis
The project includes a client that executes an external `npx` command (`@openbnb/mcp-server-airbnb`), which downloads and runs remote code. This is a critical security risk as it can introduce arbitrary, untrusted code into the execution environment. Additionally, one server (`mcp-server-deepdive-functionality/screenshot.py`) uses `pyautogui` to capture screenshots, posing a significant privacy risk if the server is compromised or misused. While `OPENAI_API_KEY` is handled via `.env` files (good practice), the inherent risks from `npx` and `pyautogui` are substantial.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
Cloudmeru icon
Sec8

Provides read-only access to MetaTrader 5 market data, technical analysis, and forecasting capabilities via Model Context Protocol.

Setup Requirements

  • ⚠️Requires Windows OS as the MetaTrader5 library is Windows-only.
  • ⚠️MetaTrader 5 terminal must be installed and running on the same machine.
  • ⚠️Python 3.10+ is required.
  • ⚠️Algorithmic trading must be enabled within the MetaTrader 5 terminal (Tools → Options → Expert Advisors).
Verified SafeView Analysis
The `execute_mt5` tool allows arbitrary Python code execution. However, the server implements robust guardrails: a restricted namespace whitelisting only read-only MT5 functions and data science libraries, explicit blocking of dangerous functions (`mt5.initialize()`, `os.system`, `eval`, `exec`, `subprocess`), command length limits, and IP-based rate limiting for HTTP. While not entirely risk-free due to code execution, these mitigations significantly enhance safety within its intended operational scope.
Updated: 2025-12-03GitHub→
0
0
Low Cost
aritracodespy icon
Sec3

This server acts as a remote intermediary for trend analysis, receiving analysis requests via WebSocket, forwarding them to another backend service, and relaying results back to clients, while also providing an API for data reporting.

Setup Requirements

  • ⚠️Requires a separate 'remote server' running at the `REMOTE_SERVER_URL` to perform the actual trend analysis, as this server acts primarily as a proxy/orchestrator.
  • ⚠️The data-store uses an in-memory array for simplicity, meaning all reported data is lost on server restart, making it unsuitable for persistent data storage without modifications.
  • ⚠️Lacks any authentication or authorization, making it vulnerable to unauthorized access and misuse if deployed in a production environment.
Review RequiredView Analysis
The server uses a wildcard CORS origin (`*`) for Socket.IO, which is a significant security risk for production environments. It directly forwards data received via WebSocket (`requestTrendAnalysis` event) to an upstream `REMOTE_SERVER_URL` without clear input validation, potentially exposing it to Server-Side Request Forgery (SSRF) or other injection attacks if the upstream service is vulnerable or the data contains malicious payloads. Critically, there are no visible authentication or authorization mechanisms for either the WebSocket connections or the `/api/report-data` endpoint, meaning any client can connect, request analysis, and report data, making it an open relay.
Updated: 2026-01-17GitHub→
PreviousPage 631 of 760Next