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

mcp-proxy

by pratikjadhav2726

Sec6

Optimizes MCP client-to-server communication by applying field projection and grep search to tool responses, reducing token usage and enhancing privacy.

Setup Requirements

⚠️Requires Python 3.12 or later.
⚠️Requires `uv` or `pip` for dependency management.
⚠️A `config.yaml` file must be present to configure underlying MCP servers, which the proxy then spawns (e.g., via `npx` or `python` commands).

Verified SafeView Analysis

The proxy handles underlying server commands and arguments from a local `config.yaml`. If this configuration file is compromised, it could lead to arbitrary command execution on the host machine. A more direct client-side risk is a potential Regular Expression Denial of Service (ReDoS) vulnerability in the `GrepProcessor`. The regex `pattern` for grep is directly sourced from client arguments, and while `maxPatternLength` is declared (1000 characters), it does not fully mitigate ReDoS attacks. There is no explicit timeout on regex compilation or matching.

Updated: 2025-12-10GitHub→

Low Cost

RubyRailsMCPServer

by PhilippGoecke

Sec6

Provides a Micro-Capability Protocol (MCP) server that exposes todo management functionalities for integration with LLM agents.

Setup Requirements

⚠️Requires a local Ollama instance to be running.
⚠️Requires Podman (or Docker) for container orchestration.
⚠️Client expects server to be accessible via specific container networking ('http://host.containers.internal:3000/mcp/sse').

Verified SafeView Analysis

The Python server uses a hardcoded 'secret-token' for its StaticTokenVerifier, which is a significant security risk for non-demo deployments. The server binds to '0.0.0.0', requiring proper firewalling in a production environment. No direct 'eval' or other arbitrary code execution patterns were observed in the provided server code snippets.

Updated: 2026-01-13GitHub→

Medium Cost

re-kb-mcp

by BadlyDrawnBoy

Sec9

A Model Context Protocol (MCP) server for managing Reverse Engineering Knowledge Base (RE KB) findings, assets, and research threads, supporting integration with LLMs and Ghidra.

Setup Requirements

⚠️Requires Python 3.11+.
⚠️Write operations (e.g., creating/updating findings, attaching evidence) require the `MCP_ENABLE_WRITES` environment variable to be explicitly set to `1`, as it defaults to `false`.
⚠️The server requires an active RE KB project to be selected before tools can be called. This can be done manually via `rekb select --root .` or automatically through session tokens (for SSE) or CWD discovery (for stdio).
⚠️The Server-Sent Events (SSE) transport (`GET /sse`) enforces a single active client connection. Subsequent connection attempts will be rejected with an HTTP 409 Conflict status until the first stream disconnects.

Verified SafeView Analysis

The server implements robust input validation using Pydantic models, transport-level connection guarding (e.g., single active SSE client, readiness gates before processing messages), and strict write controls. Write operations are disabled by default and require an explicit `MCP_ENABLE_WRITES=1` environment variable. Session tokens for SSE auto-binding are cryptographically random and validated. Cross-platform file locking (`portalocker`) is used for concurrent ID allocation to prevent race conditions and data corruption. There are no indications of 'eval', obfuscation, or hardcoded sensitive credentials, with example tokens clearly marked as such in non-production files.

Updated: 2025-12-27GitHub→

Low Cost

test-remote-mcp-server

by Bhanuraj23m0316iitb

Sec10

Provides a simple HTTP API for performing basic calculations and retrieving server metadata.

Setup Requirements

⚠️Python 3.12+ only
⚠️Requires fastmcp library

Verified SafeView Analysis

The server implements simple arithmetic and random number generation via FastMCP. No dynamic code execution, hardcoded secrets, or unusual network risks are present. It binds to 0.0.0.0, which is standard for a server intended to be accessible, but should be considered in deployment.

Updated: 2025-12-01GitHub→

Low Cost

PlaywrightMCPServer

by priyanshujain10

Sec5

Demonstrates how to use Playwright MCP server for creating automation test scripts.

Setup Requirements

⚠️Requires Node.js
⚠️Requires Playwright library installation

Review RequiredView Analysis

Only `README.md` was provided as source code. No executable code was available for security analysis. Potential risks like 'eval', obfuscation, network vulnerabilities, or hardcoded secrets could not be assessed due to the absence of functional code. The score is neutral reflecting this limitation.

Updated: 2025-11-24GitHub→

Medium Cost

PSMCP

by astrocyte

Sec9

A comprehensive WordPress management system for SST.NYC, combining MCP server tools, LearnDash LMS automation, WooCommerce integration, SEO analysis, and affiliate program management.

Setup Requirements

⚠️Requires Python 3.10+
⚠️Requires SSH access to the WordPress server with wp-cli installed.
⚠️Requires a WordPress Application Password for REST API access.
⚠️Correct configuration of WP_SSH_KEY_PATH or WP_SSH_PASSWORD is critical for SSH connection.
⚠️If Google Sheets integration is desired for affiliate management, Google Cloud OAuth credentials are required (service account or desktop app OAuth).

Verified SafeView Analysis

The Python MCP server itself is well-structured, utilizing secure libraries (paramiko for SSH, requests for HTTP/APIs) with timeouts and input validation via Pydantic/dataclasses. Remote command execution via wp-cli over SSH is an inherently powerful operation, but it is sandboxed to 'cd {remote_path} && wp {command}', and wp-cli itself handles subcommand sanitization. Input schemas for MCP tools further limit arbitrary command injection. The included PHP plugins for affiliate and class registration demonstrate strong security practices, including nonce and capability checks, robust input sanitization, secure file upload/serving (with path traversal prevention, file type/size/MIME validation, and .htaccess protection), and prepared statements for database interactions. A minor information leak risk exists where some error logs might contain PII, but no direct exploitation vector is apparent.

Updated: 2025-12-05GitHub→

Medium Cost

api-pilot

by consigcody94

Sec8

Provides an MCP server for API mocking, HTTP client requests, and OpenAPI specification exploration through natural language.

Setup Requirements

⚠️Requires Node.js 18+ and npm for local execution.
⚠️Requires Claude Desktop for Model Context Protocol (MCP) integration.
⚠️Requires manual JSON configuration in Claude Desktop's config file, specifying an absolute path to the compiled `dist/index.js`.

Verified SafeView Analysis

The project uses standard libraries (Express, Axios, SwaggerParser) without obvious code-level vulnerabilities. There are no hardcoded secrets or 'eval' usage. The `make_http_request` and `parse_openapi_spec` tools allow interaction with arbitrary URLs and local files, which is an intended feature. However, if a malicious prompt were to control these tools, it could potentially be used for Server-Side Request Forgery (SSRF) to scan internal networks or access local files. This is a risk inherent to tools that can access arbitrary external/internal resources via user input, rather than a specific vulnerability in the code.

Updated: 2025-11-20GitHub→

Low Cost

DT1_assigment_2_MCP_Server_Claude

by BajpaiH22

Sec8

Provides intelligent travel planning functionality through Claude, including real-time weather, place discovery, itinerary generation, and route optimization.

Setup Requirements

⚠️Requires Docker Desktop to be installed and running.
⚠️Requires Claude Desktop to be installed and configured.
⚠️Relies on external, free-tier APIs (Nominatim, Open-Meteo, Overpass API) which may have rate limits or occasional downtime.

Verified SafeView Analysis

The code does not use `eval` or display obfuscation. It relies on standard external APIs (Nominatim, Open-Meteo, Overpass API) which are publicly accessible and do not require API keys for this usage. Input parameters have basic type hints and some length/value constraints (e.g., `days` limited to 7, `limit` for places). All external HTTP requests include a `User-Agent` header. Error handling is present for API call failures. No hardcoded secrets or sensitive credentials were found. The server is designed to run in a Docker container, providing an isolated execution environment.

Updated: 2025-11-29GitHub→

Low Cost

mcp_server_yandex_tracker

by FractalizeR

Sec9

Integrate Yandex.Tracker API with AI clients (LLMs) for comprehensive issue, project, comment, and worklog management, enabling automated workflows and enhanced interaction.

Setup Requirements

⚠️Requires Node.js version 22.0.0 or higher.
⚠️Requires a valid Yandex OAuth token for API access (set via `YANDEX_TRACKER_TOKEN` environment variable).
⚠️Requires an Organization ID for API access; either `YANDEX_ORG_ID` (for Yandex 360) or `YANDEX_CLOUD_ORG_ID` (for Yandex Cloud Organization) must be set, but not both.

Verified SafeView Analysis

The project prioritizes security by explicitly handling secrets (API tokens, organization IDs) through environment variables only, preventing their persistence in configuration files. The `ConfigManager` is designed with a `safeFields` mechanism to ensure only non-sensitive data is saved. Development practices include `gitleaks` for secret scanning in pre-commit hooks, `dependency-cruiser` for validating dependency graphs to prevent malicious imports, `npm audit`, and `socket.dev` for supply-chain security analysis. Communication with MCP clients utilizes the standard JSON-RPC over stdio, which is generally secure. No 'eval' or obvious obfuscation patterns were observed in the provided code.

Updated: 2026-01-01GitHub→

Medium Cost

jira-mcp-server

by kushb05

Sec8

This server integrates an AI assistant with Jira for automated test case generation, ticket management, sub-task creation, and linking of related tickets.

Setup Requirements

⚠️Requires Java 23 or higher.
⚠️Requires a Jira account with an API token (Jira is a paid service).
⚠️The .env file with Jira credentials must be located in the same directory as the runnable JAR file.

Verified SafeView Analysis

The server uses environment variables (or .env file) for sensitive Jira credentials, which is a good practice to avoid hardcoding secrets. Authentication to Jira is handled via Basic Auth with an API token over HTTPS. Communication with the MCP client (Claude Desktop) occurs via stdin/stdout, which reduces direct network exposure of the server itself. No 'eval' or similar dangerous dynamic code execution patterns were found. The primary security consideration is safeguarding the Jira API token configured in the environment.

Updated: 2025-11-28GitHub→

Low Cost

MCP-Server

by eragontuan1998

Sec8

Implements a custom Minecraft Protocol (MCP) server for hosting Minecraft games.

Setup Requirements

⚠️Requires Java Runtime Environment (JRE) to run.
⚠️Requires a 'server.properties' file for server configuration.

Verified SafeView Analysis

The server relies on configuration files (server.properties) for sensitive data like host/port, which is a good practice. It connects to api.github.com over HTTPS for update checks, which introduces a minor supply-chain risk but is standard. No 'eval' or hardcoded secrets were found in the provided code snippets.

Updated: 2025-11-26GitHub→

Low Cost

mcp-hub

by DxVerm

Sec9

MCP Hub serves as an open-source client-side registry to discover, browse, and manage Model Context Protocol (MCP) servers, generating configuration snippets for Claude Code.

Setup Requirements

⚠️Custom servers and saved preferences are stored exclusively in the browser's localStorage, meaning they are not synchronized across devices and can be lost if browser data is cleared. Users should utilize the import/export functionality for backups.
⚠️The application acts as a registry and configuration generator for MCP servers; it does not host or run the MCP servers themselves. Users need to manually install and manage the actual MCP servers for Claude Code.

Verified SafeView Analysis

MCP Hub is a client-side application with no backend or server-side code execution, significantly reducing inherent security risks. It stores all user data (saved and custom servers) exclusively in browser localStorage, making it privacy-preserving as no data is sent externally. The application transparently generates configuration snippets for MCP servers, which the user then manually integrates into their Claude Code environment. The 'Security Policy' and 'CONTRIBUTING.md' explicitly highlight that users should verify the source and commands of any MCP server they configure, as MCP Hub itself does not execute these commands. The primary security consideration lies with the external MCP servers users choose to integrate, not the MCP Hub application itself. No 'eval' or similar dangerous patterns are used for arbitrary code execution within the application's core logic.

Updated: 2025-12-15GitHub→

PreviousPage 326 of 713Next