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
sahilpate3345 icon

mcp_custom_server

by sahilpate3345

Sec8

A Streamlit-based chatbot that uses MCP and LangGraph to perform general math tasks.

Setup Requirements

  • ⚠️Requires GOOGLE_API_KEY (Paid) to interact with the Gemini LLM.
  • ⚠️The MCP Server (custom_mcp_server.py) must be running locally on http://127.0.0.1:8000 before the client/web interface can connect.
  • ⚠️Requires Python environment setup with specific dependencies (e.g., mcp, langchain-mcp-adapters, langchain-google-genai, langgraph, streamlit, python-dotenv).
Verified SafeView Analysis
The server defines specific math tools (add, multiply, divide, square_root, factorial) which are inherently safe and do not allow for arbitrary code execution. Secrets (GOOGLE_API_KEY) are loaded from environment variables using `dotenv` rather than being hardcoded. The `FastMCP` server binds to a local address (127.0.0.1), limiting direct external exposure. No explicit use of dangerous functions like `eval` or `exec` is present in the provided code snippets for user input processing.
Updated: 2026-01-17GitHub→
0
0
High Cost
Sec9

Provides a secure MCP server for AI assistants to interact with IntelligenceBank APIs using OAuth 2.0 authentication and compliance review tools.

Setup Requirements

  • ⚠️Requires an IntelligenceBank account with API access and an OAuth client configured to obtain OAUTH_CLIENT_ID and platform-specific credentials.
  • ⚠️Relies on an external IntelligenceBank OAuth Bridge service (specified by OAUTH_BRIDGE_URL) for the OAuth 2.0 flow.
  • ⚠️Requires careful environment variable configuration (e.g., OAUTH_CLIENT_ID, OAUTH_REDIRECT_URI, OAUTH_BRIDGE_URL) for both local development and production.
Verified SafeView Analysis
The server explicitly prioritizes security by implementing OAuth 2.0 with PKCE, requiring HTTPS for production, configuring CORS, enabling DNS rebinding protection, and handling token expiry/refresh. It acts as an authenticated proxy to an external API. Production deployment guidelines emphasize robust security practices (EC2, nginx, SSL). The primary network risks relate to properly securing the exposed MCP endpoint and ensuring the OAuth bridge is trustworthy. No obvious 'eval' or obfuscation risks were found.
Updated: 2025-11-17GitHub→
0
0
High Cost
SAP-samples icon
Sec9

This repository provides a hands-on tutorial for agentic app development with AI clients (Cline) and SAP's MCP servers to create and modify SAP Fiori applications.

Setup Requirements

  • ⚠️Requires access to SAP Business Application Studio (SBAS).
  • ⚠️Requires API keys for AI providers (e.g., Anthropic Claude, Google Gemini) which may incur costs and need to be obtained and configured.
  • ⚠️Relies on `npx` to download and run MCP server packages, requiring a Node.js environment and network access.
Verified SafeView Analysis
The source code primarily consists of documentation (README files) and configuration for an AI client (Cline) to interact with SAP's Managed Code Project (MCP) servers. It guides users to run official SAP packages (`@sap-ux/fiori-mcp-server`, `@cap-js/mcp-server`) via `npx`. There are no 'eval' statements, obfuscation, or overtly malicious patterns in the provided snippets. The tutorial specifies using 'dummy' credentials for local application previews, which is acceptable for a development hands-on but would be a critical risk in production. The overall security depends on the user's secure handling of API keys for AI models and the security posture of the SAP Business Application Studio environment itself.
Updated: 2025-12-09GitHub→
0
0
Low Cost

ratatoskr-mcp-server

by sramkrishna

Sec9

A local MCP server for GNOME desktop integration, offering desktop management, AI-powered image analysis, face recognition, and file operations.

Setup Requirements

  • ⚠️Python 3.9-3.13 only (3.14+ not supported)
  • ⚠️Requires GNOME desktop environment for core features
  • ⚠️Podman required for AI image analysis and a ~4.3GB model download
Verified SafeView Analysis
The project emphasizes local, on-device processing for all AI features ('No Cloud APIs'). It uses standard local IPC (D-Bus) for GNOME integration and stores all data locally. File operations are explicitly 'safe' with limits (e.g., max 10 files for trash), minimizing risk of accidental large-scale damage. Podman provides container isolation for the LLaVA model. The primary risks would stem from granting it excessive local file system access or running it in an insecure local environment.
Updated: 2025-11-16GitHub→
0
0
High Cost
lanonasis icon

onasis-mcp-server

by lanonasis

Sec2

Provides a unified API gateway and Model Context Protocol (MCP) server for enterprise memory management (Memory as a Service - MaaS), AI agent integration, and secure API key management, with real-time updates and workflow orchestration capabilities.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid) for memory embeddings and AI workflow orchestration.
  • ⚠️Requires a Supabase account (PostgreSQL with vector extensions enabled) for persistent storage and database functions.
  • ⚠️Requires a Redis instance for OAuth state management and API key caching.
  • ⚠️Critical security defaults are present (e.g., default JWT_SECRET, EMERGENCY_BOOTSTRAP_TOKEN) that MUST be changed before any serious deployment.
  • ⚠️Complex environment variable configuration is necessary for full functionality, spanning database, AI, caching, and several internal service URLs.
Review RequiredView Analysis
The server has critical security vulnerabilities. The '2025-08-23-PROGRESS-UPDATE.md' explicitly flags 'Authentication & Authorization Issues ⚠️' noting 'Current MCP implementation bypasses Core authentication entirely' and 'MCP calls database directly without proper JWT validation'. The `src/netlify/mcp.js` Netlify function hardcodes `organizationId: 'ADMIN_ORG'` and `userId: null`, allowing unauthenticated creation of memories and API keys. The `src/middleware/_middleware.js` includes a 'WARNING: Placeholder implementation - DO NOT DEPLOY TO PRODUCTION' for JWT validation. The `src/routes/emergency-admin.ts` creates an admin API key without login and warns to 'Remove this file after initial setup!'. Weak default `JWT_SECRET` and `EMERGENCY_BOOTSTRAP_TOKEN` values are present. The `EncryptionUtils` uses a hardcoded 'salt' which is a security anti-pattern. While `src/middleware/auth-aligned.ts` attempts to fix some authentication issues, the overall system, especially in its Netlify function deployments, is highly insecure and not safe for production as-is.
Updated: 2026-01-02GitHub→
0
0
Low Cost
raffaeler icon

PyMcp

by raffaeler

Sec8

Demonstrates a sample Model Context Protocol (MCP) server with basic text processing and AI integration using the Python FastMCP library.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️The `oneLiner` tool uses `context.sample()` which necessitates an underlying LLM configured (e.g., an OpenAI API key or local LLM setup) for full functionality.
  • ⚠️The `uv` package manager is strongly suggested for dependency management.
Verified SafeView Analysis
The server's code does not contain 'eval', obfuscation, or hardcoded secrets. It uses `os.getenv` for configuration (host/port) and `python-dotenv` for loading environment variables. The `oneLiner` tool utilizes `context.sample()` to interact with an external AI model. This interaction implicitly relies on an LLM setup (e.g., API keys), and the `include_context='thisServer'` parameter means this server's tools could be called by the LLM. While this is part of FastMCP's design, it requires awareness regarding potential prompt injection if user inputs are not adequately handled by the underlying LLM system, though no direct vulnerability is observed in the provided server code.
Updated: 2025-11-20GitHub→
0
0
Medium Cost
kurojs icon
Sec8

Integrates ElevenLabs Text-to-Speech with OpenCode, allowing an AI assistant to speak its responses aloud through a Model Context Protocol (MCP) server.

Setup Requirements

  • ⚠️Requires ElevenLabs API Key (paid service)
  • ⚠️Automatic audio playback requires PulseAudio (Linux only)
  • ⚠️Requires OpenCode with MCP support for full integration
Verified SafeView Analysis
The server retrieves the ElevenLabs API key from an environment variable (process.env.ELEVENLABS_API_KEY), which is a good security practice. It uses `child_process.execAsync` to play audio via `paplay` on Linux, which always introduces a potential risk, but the temporary filename is generated internally using `Date.now()` and is properly quoted, mitigating command injection risks from user input. User-provided text is sent to the ElevenLabs API, not directly to `execAsync`. No `eval` or obvious malicious patterns were found. The primary external interaction is with the ElevenLabs API over HTTPS.
Updated: 2025-12-17GitHub→
0
0
Low Cost
hitosh308 icon

mcp-servers

by hitosh308

Sec9

A minimal Japanese web interface for listing and filtering Model Context Protocol (MCP) servers from a live registry, with a local data fallback.

Setup Requirements

  • ⚠️PHP is required to run the server.
Verified SafeView Analysis
The code demonstrates good security practices for a simple web application. All output to HTML is properly escaped using `htmlspecialchars`, mitigating Cross-Site Scripting (XSS) risks. External API calls are made to a hardcoded public registry URL, without dynamic user input in the URL, reducing direct server-side injection vectors. No `eval()` or similar dangerous functions are used. There are no hardcoded secrets or sensitive credentials. Client-side filtering in JavaScript further limits server-side exposure to user input for search parameters. The primary external interaction is with the public MCP registry.
Updated: 2025-12-14GitHub→
0
0
Medium Cost
DelicateAlchemy icon

loist-mcp-server

by DelicateAlchemy

Sec9

Processes audio files from URLs, extracts comprehensive metadata, stores them in Google Cloud Storage, and provides a searchable library via a Music Library Protocol (MCP) API.

Setup Requirements

  • ⚠️Requires PostgreSQL database instance (local or Cloud SQL).
  • ⚠️Requires Google Cloud Storage bucket and appropriate service account permissions.
  • ⚠️Requires FFmpeg binary installed and accessible in system PATH.
  • ⚠️Requires ExifTool binary installed and accessible in system PATH.
Verified SafeView Analysis
The server demonstrates strong security practices including comprehensive input validation (Pydantic schemas), robust URL/SSRF protection with explicit blocking of dangerous protocols and private IP ranges, and thorough SQL injection prevention via parameterized queries and psycopg2.sql. Dynamic SQL generation is handled safely. Temporary files and cleanup are managed. Potential minor risks include implicit Jinja2 auto-escaping (though usually safe by default) for metadata rendering in embed pages, and reliance on external binaries (FFmpeg, ExifTool) requiring trust in their security. Authentication uses a simple, configurable bearer token, which is noted as a limitation for production but adequate for MVP.
Updated: 2025-12-22GitHub→
0
0
Medium Cost
Sec5

A multi-country platform server designed to integrate with and manage EasyPost shipping logistics across various geographical regions.

Setup Requirements

  • ⚠️Requires EasyPost API Key (Account required)
  • ⚠️Requires a specific Python version (defined in .python-version)
Review RequiredView Analysis
Cannot fully assess without code review; server likely handles sensitive shipping data and involves network communication with third-party APIs.
Updated: 2025-11-20GitHub→
0
0
Medium Cost
enzokamal icon

mcp-hubspot-server

by enzokamal

Sec8

Provides a Model Context Protocol (MCP) server for interacting with the HubSpot CRM API, allowing AI agents to manage companies, contacts, leads, engagements, associations, and products.

Setup Requirements

  • ⚠️Requires `HUBSPOT_ACCESS_TOKEN` environment variable.
  • ⚠️Requires Node.js 18+.
  • ⚠️Uses `pnpm` for package management (though `npm` might also work).
Verified SafeView Analysis
The server securely handles API requests to HubSpot, using `HUBSPOT_ACCESS_TOKEN` from environment variables. Input parameters for tools are validated using Zod schemas, mitigating common injection risks against the HubSpot API. Error handling is present to prevent exposing raw stack traces. The primary security consideration is the secure management of the `HUBSPOT_ACCESS_TOKEN` in the deployment environment.
Updated: 2025-12-11GitHub→
0
0
Medium Cost
mhosseinkarimi icon

Chatbot_Agent

by mhosseinkarimi

Sec8

A chatbot agent leveraging large language models to interact with users and perform web scraping, crawling, and data extraction tasks via the Firecrawl service.

Setup Requirements

  • ⚠️Requires Node.js (version 16 or higher) and npx to be installed and in the system PATH.
  • ⚠️Requires an OPENAI_API_KEY environment variable (OpenAI is a paid service).
  • ⚠️Requires a FIRECRAWL_API_KEY environment variable.
  • ⚠️Requires Python 3.12 or newer.
Verified SafeView Analysis
The Python code avoids direct use of dangerous functions like 'eval' or 'exec' on user input. Subprocess calls are to fixed commands ('node --version', 'npx firecrawl-mcp') rather than user-controlled input, mitigating direct command injection risks within the Python layer. API keys are loaded from environment variables, which is good practice. The primary security consideration lies with the external 'firecrawl-mcp' tool itself, which is executed via npx. The agent's ability to call this tool implies trust in the 'firecrawl-mcp' package to handle inputs securely and perform its functions without introducing vulnerabilities that could be exploited by malicious or malformed agent-generated tool calls.
Updated: 2025-11-20GitHub→
PreviousPage 292 of 760Next