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
Medium Cost
hildersantos icon

bike-mcp-server

by hildersantos

Sec7

This server allows AI assistants to programmatically interact with the Bike outliner app on macOS, enabling reading and manipulation of outline documents.

Setup Requirements

  • ⚠️Requires macOS operating system
  • ⚠️Requires Node.js 18+ installed
  • ⚠️Requires the Bike app to be installed and running for operations
Verified SafeView Analysis
The server utilizes `child_process.execSync` to execute AppleScript commands, which inherently carries a risk of command injection if inputs are not perfectly sanitized. The code attempts to mitigate this by escaping user-provided strings for AppleScript literals (e.g., handling quotes, newlines, backslashes) and validating row IDs with a regex. However, the complexity of AppleScript escaping and the direct use of `osascript` in a shell context means a subtle vulnerability in the escaping logic or the Bike app's AppleScript interface could potentially lead to local command execution. Given it's a local-only server and relies on specific app integration, the immediate remote attack surface is limited, but local privilege escalation or unintended actions are a possibility.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
doveretepergkhb icon

arxiv-mcp-server

by doveretepergkhb

Sec5

This server provides an MCP interface for AI assistants to search, retrieve, and locally manage arXiv research papers.

Setup Requirements

  • ⚠️Requires Python dependencies from `requirements.txt`.
  • ⚠️Configuration is managed via `config/settings.example.json`.
  • ⚠️Requires an authentication header for MCP clients to connect.
Review RequiredView Analysis
Cannot perform a full security audit as only the `README.md` was provided, not the actual server source code files (e.g., Python scripts). The `README` mentions an 'authentication header' for MCP clients, implying some security measure, but its implementation details cannot be verified. Without the source code, it's impossible to check for direct risks like `eval` usage, obfuscation, hardcoded secrets, or improper handling of network requests. The project involves external network requests to arXiv, which inherently carries some risk if not handled securely.
Updated: 2025-12-11GitHub→
0
0
Medium Cost
ssuyjk-del icon

my-mcp-server

by ssuyjk-del

Sec9

This MCP server provides a boilerplate for developing AI agent capabilities (tools, resources, prompts) in TypeScript, including functions for basic utilities, image generation via Hugging Face, and a code review prompt.

Setup Requirements

  • ⚠️Requires Node.js environment and npm.
  • ⚠️Requires Smithery CLI for build/development scripts.
  • ⚠️Requires Hugging Face API token (HF_TOKEN) for the image generation tool; while a free tier exists, extensive usage may incur cost.
Verified SafeView Analysis
The server uses Zod for input validation, which is a good practice for preventing schema-related vulnerabilities. Secrets (Hugging Face API token) are managed via environment variables or configuration schema, not hardcoded. The temporary suppression of stdout during image generation is unusual but contained with a `try...finally` block. Standard network security practices should be applied when deploying the server.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
donggongailv icon

weifangbus-mcp-server

by donggongailv

Sec3

Acts as a Model Context Protocol (MCP) server providing an API wrapper for Weifang public bus real-time data and user authentication.

Setup Requirements

  • ⚠️Requires Java Development Kit (JDK)
  • ⚠️Dependent on external Weifang bus APIs (mobile.wfsmk.cn and 122.4.254.30)
  • ⚠️Requires a built JAR file (e.g., via Gradle or Maven build)
Review RequiredView Analysis
Critical cryptographic keys (APP_KEY, SERVER_PRIVATE_KEY, SERVER_DG_KEY, MCHNTID) are hardcoded, which severely compromises the security of all cryptographic operations (SM2, SM3, SM4, HMAC-SHA256) if the codebase is exposed. The bus service client uses an insecure HTTP endpoint (http://122.4.254.30:8088), making traffic susceptible to eavesdropping and tampering. A function named `encryptedString` implements a simple character shift (Caesar cipher-like) and is used for parameters like RouteID and StationID, providing very weak security for these values.
Updated: 2025-12-12GitHub→
0
0
Low Cost
echapmanFromBunnings icon

mcp.discovery.tool

by echapmanFromBunnings

Sec8

Automated discovery and security analysis of Model Context Protocol (MCP) server capabilities in .NET assemblies.

Setup Requirements

  • ⚠️Requires .NET 10.0 SDK or later
Verified SafeView Analysis
The tool acts as a security scanner for other .NET assemblies, detecting various vulnerabilities like hardcoded secrets, prompt injection, and dangerous operations. It uses isolated `AssemblyLoadContext` for safe assembly loading, mitigating risks when scanning target binaries. No 'eval' or direct malicious patterns are present in the tool's source. The primary risks would come from vulnerabilities in the assemblies being scanned, which the tool aims to identify.
Updated: 2025-12-23GitHub→
0
0
Low Cost
IT-Square-Plus icon

CLAUDE.md

by IT-Square-Plus

Sec9

Optimizes Claude Code CLI's behavior by providing structured, research-backed instructions and dynamic configuration for Model Context Protocol (MCP) servers.

Setup Requirements

  • ⚠️Requires Claude Code CLI to be installed and used.
  • ⚠️Full functionality depends on external MCP servers (e.g., Yggdrasil, Context7, Serena) being configured and accessible via `.mcp.json`.
  • ⚠️Requires manual creation of a `meta.yaml` file in the project root for consistent Yggdrasil tagging and project identification.
Verified SafeView Analysis
The project consists of an LLM configuration file and documentation, not executable code that runs a server itself. It focuses on guiding an LLM's behavior with strong safeguards, like explicitly requiring user permission for all Git operations and preventing unsolicited code changes. The main risks would be related to the LLM's interpretation of instructions or the security of external MCP servers it's configured to interact with, not vulnerabilities within the `CLAUDE.md` file itself. No 'eval' or malicious patterns are present.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
Tributary-ai-services icon

tas-mcp-servers

by Tributary-ai-services

Sec8

A collection of pre-built Model Context Protocol (MCP) servers providing capabilities like search, web scraping, database access, and development tools for local deployment and integration with the TAS MCP federation platform.

Setup Requirements

  • ⚠️Docker is required for containerized deployment or TAS MCP integration.
  • ⚠️Node.js and npm are required for local development and running individual servers directly.
  • ⚠️External API keys (e.g., BRAVE_API_KEY, Apify key) will be required for specific server functionalities that integrate with external services.
Verified SafeView Analysis
The documentation indicates good security practices for secrets management by using environment variables (e.g., BRAVE_API_KEY). It also explicitly mentions 'safety controls' for potentially sensitive operations like 'File System MCP' and 'query controls' for 'PostgreSQL MCP.' However, without the actual source code for these server implementations, a thorough audit for vulnerabilities like 'eval' or direct malicious patterns is not possible. The description itself does not contain suspicious patterns or hardcoded secrets.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
ldroguetti icon

saturn-mcp-server

by ldroguetti

Sec7

To search and retrieve documentation articles from Rocketbot Saturn for use with an MCP client.

Setup Requirements

  • ⚠️Requires Python 3.x and pip for local setup.
  • ⚠️Requires Docker and Docker Compose for containerized setup.
  • ⚠️Relies on external service `docs.rocketbot.com` for content.
Verified SafeView Analysis
The server uses `requests` to fetch content from `docs.rocketbot.com`. The `read_doc` tool includes a URL validation (`startswith(BASE_URL)`) to mitigate arbitrary Server-Side Request Forgery (SSRF) to external hosts, limiting fetches to the specified documentation domain. The `BeautifulSoup` parser is generally safe but relies on the integrity of the remote content. No `eval`, command injection, or obvious hardcoded secrets were found. The server exposes an HTTP endpoint (port 8000) which requires appropriate network security if exposed publicly.
Updated: 2025-11-25GitHub→
0
0
Low Cost
ifmelate icon

clarify-mcp

by ifmelate

Sec9

Enables AI agents to ask clarification questions and receive structured user input through a human-in-the-loop interface via the Model Context Protocol.

Setup Requirements

  • ⚠️Requires Python 3.8 or higher.
  • ⚠️Requires an MCP-enabled client (e.g., Cursor, Claude Desktop, VS Code with MCP extension) for interaction.
  • ⚠️Client configuration needs the absolute path to the 'hitl_server.py' file.
Verified SafeView Analysis
The server is explicitly designed for local stdio transport and strongly warns against network exposure without proper authentication/encryption. No malicious patterns, hardcoded secrets, or dangerous functions like 'eval' were found. Input parsing for 'ask_clarification' is robust for its purpose. Security relies on underlying FastMCP and Pydantic libraries.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
General-Mudkip icon

d2l-mcp-server

by General-Mudkip

Sec6

Provides AI assistants with tools to interact with D2L Brightspace LMS for managing academic activities.

Setup Requirements

  • ⚠️Requires manual browser interaction for initial authentication via 'd2l-mcp-auth' to save a persistent session.
  • ⚠️Requires Chromium browser installation (handled automatically by Playwright's postinstall script).
  • ⚠️Requires setting D2L_HOST environment variable for your Brightspace instance, and optionally D2L_COURSE_ID for a default course.
Verified SafeView Analysis
The server uses Playwright to manage persistent browser sessions, saving sensitive session data (cookies, local storage) to '~/.d2l-session/'. A compromise of the local machine could lead to session hijacking. The 'download_file' tool allows specifying a 'savePath', which, if controlled by a malicious or poorly designed agent, could potentially lead to arbitrary file writes on the local filesystem. However, direct external arbitrary input is unlikely to control this path as it's passed from the agent. No direct 'eval' or obvious remote code execution vulnerabilities were found.
Updated: 2025-11-28GitHub→
0
0
High Cost
tmdcpro icon

ai-pow-wow

by tmdcpro

Sec8

A collaborative AI code review system with a baseball team-themed dashboard for visualizing and managing AI agents, facilitating group discussions, and refining prompts.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid)
  • ⚠️Node.js 18+ required
Verified SafeView Analysis
Uses `process.env.OPENAI_API_KEY` for API key management, which is good practice. Input arguments for tools are parsed with `zod.safeParse`, providing validation. CORS is set to `*` for local development in `dashboard.ts`, which should be restricted for production deployments. Prompts are constructed using user-provided code, which could lead to prompt injection if not carefully managed in the LLM context.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
seanshin0214 icon

gpt-qualmaster-mcp

by seanshin0214

Sec9

Provides AI-powered tools for qualitative research methodologies and conceptual paper writing, integrating with ChatGPT Desktop via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires Python 3.9+.
  • ⚠️Dependencies must be installed via 'pip install -r requirements.txt'.
  • ⚠️A local vector database needs to be initialized by running 'python init_vectordb.py' before the server can utilize RAG features.
Verified SafeView Analysis
The server's tool handling dispatches to predefined Python functions, mitigating risks from arbitrary code execution. Data storage uses a local ChromaDB PersistentClient. No hardcoded secrets or direct system command injections from user input were observed. The CORS middleware is open, but this is typical for a local desktop companion app that may be proxied.
Updated: 2025-12-07GitHub→
PreviousPage 385 of 713Next