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
Samuel88 icon

mcp-server-test

by Samuel88

Sec8

This server provides a Model Context Protocol (MCP) endpoint to expose an addition tool and a greeting resource via HTTP.

Setup Requirements

  • ⚠️Node.js 18+ required
  • ⚠️npm or yarn required for dependency installation
Verified SafeView Analysis
The server uses Zod for schema validation which improves input safety. However, the CORS configuration (`origin: '*'`) allows requests from any domain, which might be overly permissive for production environments. The 'greeting' resource directly interpolates user-provided 'name' into its output, which, if rendered directly by a client-side application without sanitization, could lead to Cross-Site Scripting (XSS) vulnerabilities. No hardcoded secrets or 'eval' usage found.
Updated: 2025-11-25GitHub→
0
0
Low Cost
ServiceNow icon

saai-skill-feedback

by ServiceNow

Sec8

Submit feedback about Claude Model Context Protocol (MCP) skills directly from conversations, creating ServiceNow SBOs for tracking.

Setup Requirements

  • ⚠️Requires Node.js 18+ and Python 3.8+ installed locally.
  • ⚠️Requires a ServiceNow account on `surf.service-now.com` with permissions to create SBOs in the `x_snc_security_d_0_dsrtable` table.
  • ⚠️Requires interactive browser-based authentication (Okta, MFA) during installation and potential re-authentication for session refresh, leveraging `selenium` and `webdriver-manager`. Safari automation requires `sudo safaridriver --enable`.
Verified SafeView Analysis
The server uses browser automation (Selenium) for initial ServiceNow authentication, caching session cookies and an X-UserToken locally in `~/.servicenow_surf_session.json`. While this is a common practice for internal tools, the security relies on the integrity of the local file and browser environment. Credentials can also be provided via environment variables. All network communication to ServiceNow is over HTTPS. `child_process.spawn` is used to execute Python scripts, but arguments are internally controlled and not directly user-provided, mitigating injection risks. No hardcoded sensitive secrets were found.
Updated: 2026-01-16GitHub→
0
0
Medium Cost
aoki-collabo icon

simple-hono-mcp-server

by aoki-collabo

Sec8

This server provides a simple Model Context Protocol (MCP) endpoint that offers a 'factorize' tool for positive integer factorization, primarily serving as a demonstration and educational example for deploying MCP servers on Cloudflare Workers using the Hono framework.

Setup Requirements

  • ⚠️Requires Node.js v18 or higher.
  • ⚠️A Cloudflare account is needed for deployment to Cloudflare Workers.
Verified SafeView Analysis
The server explicitly lacks authentication, a documented design choice for simplicity in a demo. This means any client can call the exposed tools, making it unsafe for production environments without adding robust authentication. However, input validation for the 'factorize' tool is implemented using zod, checking for integer type, positive value, and a maximum of 15 digits, which mitigates computational DoS attacks for the tool's logic.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
1bluebird777 icon

manus-mcp-server

by 1bluebird777

Sec1

MCP server facilitating real-time development tasks for ElevenLabs' Leiah AI agent on the BluebirdX project, including booking and code interaction.

Setup Requirements

  • ⚠️Requires a specific directory structure on the host machine with a 'bluebird-x' project repository located at '/home/ubuntu/bluebird-x' for development tools to function correctly.
  • ⚠️Relies on an external 'BluebirdX API' (URL configurable via BLUEBIRD_API_URL) for booking and address validation, which must be running and accessible.
Review RequiredView Analysis
CRITICAL VULNERABILITIES: The `get_code_context` tool uses `execSync` with `grep -r "${query}"`. The `query` argument comes directly from the AI agent's input, making the server highly vulnerable to shell injection attacks. An attacker controlling the AI agent's prompts could execute arbitrary commands on the host system. Additionally, the `INTEGRATION_GUIDE.md` explicitly states 'No authentication' and 'No rate limiting' are implemented in the current setup, making the server openly accessible and susceptible to abuse. The `create_task` tool writes user-provided content (`title`, `description`) to files, which could potentially lead to path traversal or denial-of-service via large files if inputs are not properly sanitized. The `BLUEBIRD_API_URL` being configurable via an environment variable could expose the server to Server-Side Request Forgery (SSRF) if the variable is compromised. This server is not safe to run in its current state without significant security enhancements.
Updated: 2025-12-14GitHub→
0
0
Low Cost
chew-z icon

TimeMCP

by chew-z

Sec9

Provides a Model Control Protocol (MCP) server for current time retrieval and timezone conversions.

Setup Requirements

  • ⚠️Requires a Go (GoLang) development environment to build and run from source.
  • ⚠️If HTTP transport with authentication is enabled, the `TIME_AUTH_SECRET_KEY` environment variable must be set (recommended >=32 characters).
  • ⚠️Relies on the system's IANA timezone database for timezone operations.
Verified SafeView Analysis
The server demonstrates strong security practices for network exposure, including explicit CORS configuration with secure defaults (disabled by default, allowlist required, wildcard '*' rejected with auth enabled) and robust JWT authentication with checks for secret key presence/length and proper token validation. No 'eval' or similar dangerous patterns are present. `TIME_AUTH_SECRET_KEY` is loaded from environment variables. Overall, the project prioritizes security in its design.
Updated: 2025-11-19GitHub→
0
0
Low Cost
Digital-Defiance icon

mcp-core

by Digital-Defiance

Sec8

Provides shared infrastructure for building AI capability servers that expose a set of tools.

Setup Requirements

  • ⚠️Requires Node.js environment
  • ⚠️Requires TypeScript compilation (`yarn build`)
  • ⚠️Primarily uses 'yarn' as package manager (though npm/pnpm might work)
Verified SafeView Analysis
The server-base library itself does not contain obvious critical vulnerabilities like 'eval' or hardcoded secrets. It communicates over standard I/O (StdioServerTransport), reducing direct network attack surface. Tool input validation is facilitated by Zod schemas, but the ultimate security depends on how tool handlers are implemented by the consumer, as 'request.params.arguments' is passed directly to the handler.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
8resd8 icon

Dev-Assistant

by 8resd8

Sec7

An AI-powered local server that allows an LLM to query a local MySQL database and analyze local server log files via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires a running local MySQL instance.
  • ⚠️Requires configuration of database connection details in `application.properties`.
  • ⚠️The `server.log` file must be present in the application's working directory for log analysis to function.
  • ⚠️Requires an LLM client (e.g., Claude Desktop) that supports Model Context Protocol (MCP) and needs to be configured to connect to this server.
Verified SafeView Analysis
The server restricts database queries to only 'SELECT' statements, which is a strong measure against data manipulation. It uses local Stdio for communication, preventing external network exposure. However, the `DatabaseMcpService` directly executes LLM-generated SQL queries via `jdbcTemplate.queryForList(sql)` without parameterized statements, making it potentially vulnerable to SQL injection if an LLM crafts a malicious 'SELECT' query (e.g., to exfiltrate data or cause performance issues). Log file analysis is limited to 50 lines, reducing potential resource exhaustion.
Updated: 2025-12-02GitHub→
0
0
Low Cost
BlockRunAI icon

awesome-devops-mcp

by BlockRunAI

Sec10

A curated list of Model Context Protocol (MCP) servers for DevOps, cloud infrastructure, and platform engineering, enabling AI agents to interact with various tools and services.

Verified SafeView Analysis
The provided source code consists of markdown documentation (README.md and CONTRIBUTING.md) and does not contain executable code. Therefore, there are no direct security risks such as 'eval', obfuscation, network risks, hardcoded secrets, or malicious patterns within this repository's files. The security of any *listed* MCP servers would need to be assessed independently.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
dennisonbertram icon

mcp-etherscan-server

by dennisonbertram

Sec9

Provides a multi-chain Etherscan-based interface for fetching Ethereum blockchain data, including balances, transactions, token transfers, contract details, and network statistics across 70+ networks.

Setup Requirements

  • ⚠️Requires an Etherscan API Key (free tier with rate limits, paid for higher usage).
  • ⚠️Requires Node.js >= 18.
  • ⚠️Needs an '.env' file configured with ETHERSCAN_API_KEY in the root directory.
Verified SafeView Analysis
The server correctly handles sensitive API keys via environment variables (ETHERSCAN_API_KEY is explicitly required and checked). It uses reputable external APIs (Etherscan, eth.llamarpc.com for ENS) and robust URL building with parameter encoding. No 'eval' or obvious malicious code patterns were found. Dependency usage is for standard utilities and cryptographic operations. The `parseApiResponse` function includes basic structural and status validation, reducing the risk of malformed API responses causing unexpected behavior.
Updated: 2025-12-03GitHub→
0
0
High Cost
GlacierEQ icon
Sec8

A sophisticated, AI-driven framework for multi-engine browser automation, web scraping, and cognitive task execution on the web, capable of complex problem-solving and error recovery.

Setup Requirements

  • ⚠️Requires OpenAI API Key or similar for LLM provider (Paid Service)
  • ⚠️Requires browser drivers to be installed (e.g., playwright install)
  • ⚠️Python 3.11+ required
  • ⚠️Optional: Redis or another memory backend for cognitive persistence
Verified SafeView Analysis
The framework's architecture outlines security best practices like not logging credentials, rotating proxies, and encrypting cookies. However, direct code for handling LLM API keys or external service credentials is not shown, which could be a risk if not implemented securely (e.g., hardcoded keys). Reliance on external LLMs and CAPTCHA services introduces data privacy and security considerations that depend on third-party service policies and the implementation details of data handling. No explicit 'eval' or malicious patterns are visible in the truncated source.
Updated: 2025-11-23GitHub→
0
0
Low Cost
Sec8

Enables AI assistants like GitHub Copilot to execute Power Automate Desktop flows and automate Windows UI operations.

Setup Requirements

  • ⚠️Requires Windows 10 or Windows 11.
  • ⚠️Requires Power Automate Desktop (PAD) to be installed and running, with the Console Host initialized at least once.
  • ⚠️For development, .NET 8.0 SDK is required; for release, .NET 8.0 Runtime is implicitly needed (often bundled with PAD).
Verified SafeView Analysis
The project demonstrates a strong focus on security, providing a detailed SECURITY.md that outlines input validation for flow names, window titles, and keystroke sequences, and enforcing code analysis rules. Operations are local-only and run within the current user's context, mitigating remote attack vectors. However, it relies on `child_process.exec` for system commands (`tasklist`, `powershell`), which are potential injection points if validation fails or is bypassed. Furthermore, Power Automate Desktop flows execute with full PAD permissions and are not sandboxed by the MCP server, posing a risk if a malicious flow is executed. Users must validate AI requests and manage PAD flows carefully, as highlighted in the 'Known Security Considerations'.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
Supplyline icon

plytix-mcp

by Supplyline

Sec9

Provides AI assistants with live access to Plytix PIM data via the Model Context Protocol (MCP) for searching, looking up, and retrieving product information.

Setup Requirements

  • ⚠️Requires Node.js 18+ (20+ recommended).
  • ⚠️Requires a Plytix PIM account with API access (API Key and Password).
Verified SafeView Analysis
The server demonstrates strong security practices. It avoids hardcoded secrets by relying on environment variables or request headers for API credentials. Automatic token refresh and rate limit handling with exponential backoff are implemented for robust API interaction. Input validation for tool arguments is performed using `zod`, which helps prevent common injection vulnerabilities. Explicit CORS configuration restricts access to allowed origins. There is no usage of `eval` or similar dangerous functions found in the provided source code, and the stateless nature inherently reduces some persistence-related risks.
Updated: 2026-01-19GitHub→
PreviousPage 660 of 713Next