Stop Searching. Start Trusting.

The curated directory of MCP servers, vetted for security, efficiency, and quality.

Browse Directory Why 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)

Low Cost

rampify-mcp

by rampify-dev

Sec8

Provides real-time SEO intelligence, site audits, Google Search Console integration, and AI-powered recommendations directly within an IDE for developers.

Setup Requirements

⚠️Requires Node.js 18 or higher.
⚠️Requires a Rampify account and API Key.
⚠️Requires the Rampify backend API to be running and accessible (defaults to https://www.rampify.dev).

Verified SafeView Analysis

The server primarily acts as an intermediary, querying a backend API and local development servers. It uses standard HTTP client libraries (axios, fetch). The `generate_schema` tool uses `dangerouslySetInnerHTML` for outputting JSON-LD snippets, which is a known React/Next.js risk but is generally acceptable for static schema data generated from trusted content. A private `get_security_context` tool exists but is explicitly excluded from npm builds, reducing its public exposure. No direct `eval` or blatant malicious patterns were found. Network interactions with configured backend and local dev servers are expected.

Updated: 2025-12-05GitHub→

High Cost

mailbox-mcp-server

by soenkenils

Sec9

Integrates mailbox.org email, calendar, and Sieve filter services with Anthropic's Claude Desktop application for secure, local data interaction.

Setup Requirements

⚠️Requires Node.js 20 or later.
⚠️Requires a mailbox.org account with IMAP and CalDAV access enabled.
⚠️Requires the Anthropic Claude Desktop application.
⚠️Strongly recommends using an App Password for mailbox.org for enhanced security.
⚠️Uses `bun` as the package manager, requiring `bun install` for setup.
⚠️Must be explicitly added to Claude Desktop's `claude_desktop_config.json` with the correct path to `dist/main.js`.
⚠️The default Sieve server port is configured to `2000`, while the mailbox.org standard is often `4190` (configurable via `MAILBOX_SIEVE_PORT`).

Verified SafeView Analysis

The server demonstrates strong security practices including comprehensive input validation (using Valibot) and HTML sanitization to prevent XSS. It defaults to TLS encryption for all network connections (IMAP, SMTP, CalDAV, Sieve) and logs warnings if insecure connections are configured. Credentials are managed via environment variables and app-specific passwords are recommended, with no persistent storage of sensitive data. A specific `sanitizeArgs` function is implemented to remove invalid UUID parameters potentially passed by Claude Desktop. Sieve script creation/update includes a validation step prior to deployment. Data processing remains local to the user's machine, minimizing external data exposure. The main dependencies are well-established libraries, and potential vulnerabilities within them would be external risks.

Updated: 2026-01-03GitHub→

Low Cost

adminmcp

by frankausberlin

Sec5

Provides a web-based administrative interface for managing one or more Minecraft servers, offering functionalities like server startup/shutdown, configuration editing, and potentially user/plugin management.

Setup Requirements

⚠️Requires Node.js runtime and npm/yarn for dependency management.
⚠️Likely requires a local or remote database (e.g., SQLite, MongoDB, PostgreSQL) for storing server configurations and user data.
⚠️Requires proper configuration of paths to Minecraft server `.jar` files and associated directories.
⚠️Requires an open network port for the web interface.

Review RequiredView Analysis

This application likely uses `child_process` to manage external Minecraft server processes, creating a significant risk of command injection if user-supplied inputs (e.g., server commands, file paths) are not thoroughly sanitized and validated. Robust authentication and authorization mechanisms are crucial to prevent unauthorized access to server management functionalities. Potential for hardcoded secrets (e.g., database credentials, RCON passwords, API keys) if environment variable usage is not strictly enforced. Thorough code review is recommended before deployment, especially in a public-facing environment.

Updated: 2025-12-01GitHub→

Medium Cost

mcp-code-vector--search

by wgawan

Sec8

Builds a semantic index of a codebase for natural language search and code context retrieval, often used as a tool for AI coding agents.

Setup Requirements

⚠️Requires Python 3.11+ if running without Docker (Docker bundles its own Python environment).
⚠️Installation of OpenVINO/Optimum packages for sentence-transformers can take several minutes due to the compilation and installation of native components.

Verified SafeView Analysis

The server design generally appears robust. It uses standard, reputable libraries (ChromaDB, Sentence Transformers). When run via Docker, it limits file system write access to only the index directory (`/vector-index`). Input validation for `query` and `filter_filepath` in `semantic_search` is handled by ChromaDB's query interface, reducing direct injection risks. No 'eval' or obvious hardcoded secrets are present. The communication for the MCP client interface occurs over standard I/O (`stdio_server`), which reduces the direct network attack surface. The primary risk would be if a malicious client (within a trusted environment) could craft highly inefficient queries leading to excessive resource consumption, but this is generally mitigated by running against your own codebase in a controlled environment.

Updated: 2025-12-14GitHub→

Low Cost

PlaywrightTS_ClaudeAgent

by LabsKraft-Ramendra

Sec9

A comprehensive Playwright-based automated testing framework for login form testing with Page Object Model (POM) pattern and Excel-driven test data.

Setup Requirements

⚠️Node.js and npm must be installed.
⚠️Playwright browser binaries need to be installed (`npx playwright install`).
⚠️Initial test data (`test_data.xlsx`) must be generated using `node scripts/create_login_testdata.js`.

Verified SafeView Analysis

This project is a test automation framework designed to run locally and interact with external web pages. It does not expose any server-side functionality or APIs to untrusted input. Local file system interactions (`exceljs`) are used for managing test data (`test_data.xlsx`), which is internal to the project and not expected to be user-provided external input. The inherent risks are similar to running any browser automation scripts: potential interaction with malicious external websites or local resource access if not properly secured, but the provided source code itself does not contain any obvious malicious patterns or `eval` usage. Test credentials for the target login form are stored in an Excel file for parameterization, which is appropriate for a testing framework and not considered a hardcoded secret in this context.

Updated: 2025-11-30GitHub→

Low Cost

my-first-mcp-server

by ankurvapte

Sec8

Manages employee annual leave balances and historical leave dates through an MCP server.

Setup Requirements

⚠️Requires Python 3.13+ (specified in pyproject.toml, which is a relatively new Python version)
⚠️Requires 'uv' package manager to run as instructed
⚠️Requires the 'mcp' framework to be installed

Verified SafeView Analysis

The code primarily uses in-memory data structures, limiting common web vulnerabilities like SQL injection. No 'eval' or similar dangerous functions are used. There are no hardcoded secrets or obvious malicious patterns. Input validation is minimal (e.g., employee_id existence, leave days vs. balance), but sufficient for an example server not handling sensitive external interactions. The lack of robust input sanitization for string fields like 'date' could be a minor concern in a production environment if combined with other vulnerabilities, but is not critical here.

Updated: 2025-12-01GitHub→

Medium Cost

gocardless-mcp

by jmceleney

Sec9

Enables AI assistants to interact with GoCardless payment data for listing, retrieving, and creating financial records.

Setup Requirements

⚠️Requires GoCardless API Access Token (potentially paid for live usage, free for sandbox).
⚠️Requires Python 3.10 or higher.

Verified SafeView Analysis

Handles GoCardless API access tokens via environment variables, which is secure. No `eval` or dynamic code execution found. Interacts with the GoCardless API exclusively through the official `gocardless-pro` client library. The README recommends read-only API tokens, which is a good security practice.

Updated: 2025-11-26GitHub→

Low Cost

til-mcp

by molefrog

Sec4

A minimal Model Context Protocol (MCP) server for storing and retrieving 'Today I Learned' memos, designed to expose tools for AI interaction.

Setup Requirements

⚠️Requires Bun runtime to be installed (leverages `bun:sqlite`).
⚠️Authentication is optional and disabled by default if `MCP_AUTH_SECRET` is not set, posing a significant security risk for public deployments.
⚠️Uses Express.js, not Hono, due to architectural incompatibilities with Mastra's Streamable HTTP transport (as noted in `BUN-MCP-SERVER.md`).

Review RequiredView Analysis

The server's authentication is disabled by default if the `MCP_AUTH_SECRET` environment variable is not set. This is a critical security risk for a public-facing deployment, as it allows unauthorized access to `listMemos`, `getMemo`, and crucially, `addMemo` (which can create or update entries). While input validation is present via Zod schemas and SQL injection is mitigated using prepared statements for SQLite queries, the lack of mandatory authentication by default and absence of rate limiting or more robust access controls significantly lowers the security posture.

Updated: 2025-11-28GitHub→

Low Cost

DevFest-Hangzhou

by Ryan041001

Sec1

A personal project or demonstration server for a workshop related to Chrome DevTools at DevFest Hangzhou 2025.

Review RequiredView Analysis

No source code beyond the README.md file was provided. Therefore, a comprehensive security audit is impossible. Running this project without reviewing its full source code is extremely risky, as it may contain malicious patterns, hardcoded secrets, or unpatched vulnerabilities. The score is 1 due to the inability to assess any actual code.

Updated: 2025-11-20GitHub→

High Cost

Noesis

by chetan0220

Sec7

Noesis is an AI-powered conversational data analysis platform designed to help data scientists and analysts explore, preprocess, and visualize datasets using natural language commands.

Setup Requirements

⚠️Requires Python 3.13+
⚠️Requires PostgreSQL database setup and credentials.
⚠️Requires MinIO object storage setup and credentials.
⚠️Requires an external LLM API key (likely Groq API Key for 'openai/gpt-oss-120b' model).
⚠️Requires a TinyURL API Key for URL shortening in visualizations.
⚠️The full application requires three separate services to run: FastAPI backend, MCP server, and Next.js frontend.

Verified SafeView Analysis

The application generally uses good security practices for database interaction (parameterized queries) and secret management (environment variables for DB, MinIO, JWT secret, TinyURL API key). However, a critical security risk for production environments is the `allow_origins=['*']` configuration in `main.py`, which enables Cross-Origin Resource Sharing from any origin. This should be restricted to known frontend origins in a production deployment. Additionally, while the LLM agent's output is filtered for sensitive context, sophisticated prompt injection could still potentially manipulate tool arguments or lead to unintended data operations if not robustly mitigated by the agent framework.

Updated: 2025-12-14GitHub→

Low Cost

TodoMCPServer

by cpateldev

Sec8

Integrates a Todo REST API with Model Context Protocol (MCP) server capabilities, enabling AI assistants to interact with todo list management.

Setup Requirements

⚠️The `ModelContextProtocol` NuGet package is listed in `TodoMCPServer.csproj` and `Program.cs`, but the documentation indicates it is 'not needed in this project,' which can cause confusion or unnecessary dependencies.
⚠️For Azure App Service deployments, the `.WithStdioServerTransport()` line in `Program.cs` must be commented out, as stdio transport is not supported in that environment, otherwise, it will throw a startup error.
⚠️The application uses an in-memory Entity Framework Core database, meaning all data resets upon each application restart, which is suitable for development but not for persistent data storage.

Verified SafeView Analysis

The project, as provided for demonstration and development, does not implement authentication or authorization by default, making it unsuitable for production environments without further hardening. It utilizes an in-memory database that resets on application restart, posing no data persistence risk. No obvious malicious patterns or hardcoded secrets are present in the provided code snippets. Optional steps for securing endpoints with JWT/OAuth are outlined.

Updated: 2026-01-19GitHub→

Low Cost

context-engine-mcp

by SumeetGDoshi

Sec9

Automates the Research → Plan → Validate workflow for AI coding tools to improve code quality and prevent rework.

Setup Requirements

⚠️Requires Node.js 18+.
⚠️Requires manual configuration of IDE's MCP settings (e.g., ~/.cursor/mcp.json), although an installer script can automate this for Cursor.
⚠️Requires restarting the IDE after configuration for changes to take effect.

Verified SafeView Analysis

The server primarily operates locally via standard I/O (stdio) and manages state in local project files (`.context-engine/workflow-state.json`) or a designated `mcpDocs/` directory. File system operations are confined to these known paths, limiting arbitrary file access. No direct use of `eval` or similar dynamic code execution of user input is evident. The `install.sh` script modifies user configuration files (`~/.cursor/mcp.json`), but this is an explicit setup step that prompts for user consent. The main security consideration would stem from the AI agent that this server guides, rather than the server's own code.

Updated: 2025-11-26GitHub→

PreviousPage 677 of 713Next