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
Low Cost
ericomack1983 icon

MCPDiscordChatBot

by ericomack1983

Sec1

A Discord chatbot that directly integrates with and queries ChatGPT to answer user questions, styled as Agent Smith.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid).
  • ⚠️Requires Discord Bot Token.
  • ⚠️The provided `discord_aibot.py` directly connects the Discord bot to ChatGPT, which deviates from the README's described architecture of `Discord Chat Bot → MCP Server (Python) → ChatGPT`. The 'MCP Server (Python)' component (e.g., a Flask server) mentioned in the README is not present in the provided source code.
Review RequiredView Analysis
CRITICAL: The `openai.Client` initialization explicitly uses `api_key="XXXXXXXX"` instead of `os.getenv('OPENAI_API_KEY')`. While `os.getenv` is commented out, the active line hardcodes a placeholder that could easily become a hardcoded secret if replaced without proper environment variable usage. This is a severe security vulnerability. The Discord token is correctly loaded from environment variables.
Updated: 2025-12-01GitHub→
0
0
Medium Cost
strato-net icon

strato-griphook

by strato-net

Sec7

An MCP (Model Context Protocol) server enabling AI agents to interact with the STRATO blockchain's DeFi ecosystem.

Setup Requirements

  • ⚠️Requires Node.js >= 18.0.0.
  • ⚠️Requires OAuth 2.0 client configuration (OAUTH_CLIENT_ID, OAUTH_CLIENT_SECRET, OPENID_DISCOVERY_URL) for login and hosted mode.
  • ⚠️Requires a running STRATO API backend accessible via STRATO_API_BASE_URL.
  • ⚠️For public hosted deployments, additional setup for Keycloak, DNS, Nginx, and SSL is required, as outlined in the project's deployment guide (external to source code).
Verified SafeView Analysis
The server transparently handles OAuth 2.0 (PKCE flow) for authentication. Credentials are saved unencrypted to `~/.griphook/credentials.json` with `0o600` permissions; users must protect their filesystem. For local operation, the HTTP server binds to `127.0.0.1` without TLS, requiring users to manually ensure HTTPS and authentication if exposed publicly. The `strato.rpc` tool acts as a JSON-RPC proxy, allowing AI agents to submit arbitrary payloads to the configured STRATO API. This powerful capability relies heavily on the security and validation of the backend STRATO API to prevent misuse or arbitrary execution.
Updated: 2026-01-16GitHub→
0
0
Low Cost
SuperPyonchiX icon

confluence_mcp_server

by SuperPyonchiX

Sec8

Provides a Model Context Protocol (MCP) server for AI agents to interact with Confluence DataCenter/Server content, manage pages, spaces, users, search using CQL, and perform bidirectional Markdown conversions.

Setup Requirements

  • ⚠️Requires Node.js 18 or higher.
  • ⚠️Primarily designed for Confluence DataCenter/Server with Basic Authentication (username/password), as stated in the main README. Although the code includes branches for token authentication, the README explicitly states Confluence Cloud is not supported in this version.
  • ⚠️Markdown conversion tools (`confluence_page_to_markdown`, `confluence_markdown_to_page`, `confluence_update_page_from_markdown`) require absolute file paths for inputs and outputs.
Verified SafeView Analysis
The codebase generally follows good security practices, utilizing environment variables for credentials and including checks for absolute file paths and directory traversal in file operations. It uses `axios` for network requests with interceptors for error handling. HTML/Markdown processing performs basic escaping. No explicit use of `eval` or similar dangerous patterns was found. While overall good, deeper dynamic analysis of external SDKs is beyond this scope.
Updated: 2026-01-16GitHub→
0
0
Low Cost
dannwaneri icon

mcp-server-worker

by dannwaneri

Sec9

A production-ready Model Context Protocol (MCP) server providing HTTP-based semantic search and intelligent search with AI-powered synthesis context using Cloudflare Workers AI and Vectorize.

Setup Requirements

  • ⚠️Requires a Cloudflare account with Workers enabled.
  • ⚠️Requires Wrangler CLI installed.
  • ⚠️Requires a Vectorize index named 'mcp-knowledge-base' to be created with 384 dimensions and cosine metric.
  • ⚠️Requires the Vectorize index to be populated with data (e.g., using the referenced 'vectorize-mcp-worker' project) before semantic search yields results.
  • ⚠️No built-in authentication for API endpoints by default, making it publicly accessible unless a custom authentication layer is added as suggested in the README.
Verified SafeView Analysis
The code acts as an HTTP adapter for the MCP protocol, directly handling 'tools/list' and 'tools/call' methods. User input for search 'query' is passed to Workers AI for embedding and then to Vectorize. The 'topK' parameter for search results is explicitly capped at 10, preventing excessive resource consumption. CORS is broadly enabled ('Access-Control-Allow-Origin: *'), which is common for public APIs but could be restricted further if needed. While the README suggests adding authentication as a production enhancement, the core worker logic itself does not implement it by default, meaning the API is publicly accessible post-deployment. No 'eval' or other obvious code injection vulnerabilities were found in the provided source.
Updated: 2025-12-05GitHub→
0
0
Medium Cost
Kirkigmenezes icon

AIDomesticCoreAIJ

by Kirkigmenezes

Sec9

AIDomesticCoreAIJ is a foundational artificial intelligence platform designed for both home and enterprise applications. It provides the power to create intelligent agents, automate tasks, and orchestrate complex multi-model systems, integrating quantum computing, computer vision, federated learning, generative AI, privacy-first IDEs, LLM-kernel IDEs, and edge execution capabilities.

Setup Requirements

  • ⚠️Requires Docker and Docker Swarm for deployment in a clustered environment.
  • ⚠️Python 3.8-3.11 is required (Python 3.11 recommended).
  • ⚠️Minimum 8GB RAM (16GB recommended) and 10GB storage.
  • ⚠️CUDA-compatible GPU is optional but recommended for accelerated inference and training.
  • ⚠️External API keys for various AI services (e.g., OpenAI, Claude, GigaChat3, KatyaAI) are required for full functionality of GenAI components.
  • ⚠️Uses API keys for authentication (`Authorization: Bearer <api_key>`).
Verified SafeView Analysis
The project demonstrates a robust commitment to security, featuring quantum-safe cryptography (Kyber, Dilithium), a Zero-Trust security model, Decentralized Identity Networks (DIDN), and automated SBOM generation with vulnerability scanning (`syft`, `grype`). It also outlines compliance with EU AI Act and GDPR, detailed security policies, bug bounty programs, and strong input validation. Minor findings include a hardcoded development password in `docker-compose.yml`, which is a common practice for local development setups but should not be used in production.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
deepakjain12345 icon

jira-utilities-mcp-server

by deepakjain12345

Sec9

Automating JIRA issue management and generating comprehensive reports for PR links, Zephyr test cases, and subtask analysis, either via an MCP server or standalone scripts.

Setup Requirements

  • ⚠️Requires JIRA_EMAIL and JIRA_API_TOKEN environment variables for authentication.
  • ⚠️Requires Docker or Podman to run the MCP server as a container.
  • ⚠️Requires manual configuration of Cursor's ~/.cursor/mcp.json file, including volume mounts for CSV output.
  • ⚠️Network access to celigo.atlassian.net is essential.
Verified SafeView Analysis
The project adheres to good security practices by strictly requiring JIRA API tokens and email to be passed via environment variables, avoiding hardcoding of secrets. It uses HTTPS for all JIRA API communications. Input validation is present for JIRA issue keys. The handling of JQL queries relies on the JIRA API's inherent sanitization. No 'eval' or obfuscation is used.
Updated: 2025-11-24GitHub→
0
0
Low Cost
KiranFarhan786 icon

mcp_server_kf

by KiranFarhan786

Sec1

This server likely implements a custom protocol for communication, possibly a game server given the 'MCP' prefix (often associated with Minecraft Protocol).

Review RequiredView Analysis
Cannot perform a security audit as the source code was not provided for analysis. Therefore, potential risks like 'eval' usage, hardcoded secrets, or malicious patterns cannot be identified or ruled out.
Updated: 2025-12-06GitHub→
0
0
Medium Cost
globle21 icon

mcp-server

by globle21

Sec7

Fast, token-efficient web content extraction for AI agents, converting websites to clean Markdown while preserving links and structure.

Setup Requirements

  • ⚠️Requires Node.js >=20.0.0.
  • ⚠️Uses Playwright for browser rendering, which might require additional system dependencies or a larger Docker image if browser binaries are not pre-installed or managed.
  • ⚠️The `read_website` tool by default does NOT respect `robots.txt` (can be overridden).
  • ⚠️Local cache directory (`.cache` by default) requires write permissions in the working directory.
Verified SafeView Analysis
The server's core function involves fetching content from arbitrary external URLs, which is an inherent network risk. It uses Playwright for JavaScript rendering, increasing the potential attack surface if a highly sophisticated exploit exists on a target page. The `cookiesFile` and `cacheDir` parameters, while used for local paths, could pose a risk if exposed to untrusted user input without strict validation, but in an MCP context, client inputs are generally more controlled. There are no obvious hardcoded secrets or clear malicious patterns.
Updated: 2025-11-24GitHub→
0
0
Medium Cost
Sec8

Provides LLM-powered "second opinions" on challenging coding issues by integrating Google Gemini with Claude Code CLI.

Setup Requirements

  • ⚠️Requires Conda for environment management and activation.
  • ⚠️A Google Gemini API Key is required (free tier available for limited use).
  • ⚠️Configuration within Claude Code CLI requires an absolute path to the server.py file.
Verified SafeView Analysis
The server interacts with the Google Gemini API over the network, which requires handling API keys. It is designed to be invoked locally by the Claude Code CLI, limiting direct public exposure.
Updated: 2025-11-23GitHub→
0
0
Low Cost
TaeYoung11 icon
Sec10

Provides a Micro-Agent Platform (MCP) tool to fetch current weather forecasts for specified geographical coordinates.

Setup Requirements

  • ⚠️Requires Python 3.12+
  • ⚠️Requires `mcp` library (Micro-Agent Platform)
Verified SafeView Analysis
No obvious security vulnerabilities, dangerous patterns (like 'eval'), hardcoded secrets, or malicious code identified in the provided truncated source code. It makes a standard HTTP GET request to a public weather API.
Updated: 2025-12-11GitHub→
0
0
Low Cost
LibraxisAI icon

rmcp_mux

by LibraxisAI

Sec8

Manage multiple Model Context Protocol (MCP) servers from a single process, multiplexing client connections, providing features like ID rewriting, caching, auto-restart, and a TUI dashboard.

Setup Requirements

  • ⚠️Requires Rust toolchain and Cargo for building from source (e.g., `cargo build`).
  • ⚠️Requires a Unix-like operating system (macOS/Linux) as it extensively uses Unix domain sockets.
  • ⚠️Reliable operation depends on a correctly configured and trusted `mux.toml` file, as it defines the commands and parameters for all managed child processes.
Verified SafeView Analysis
The source code demonstrates good practices against common vulnerabilities like path traversal by using `fs::canonicalize` for path sanitization. No hardcoded secrets or malicious patterns (e.g., `eval`) were found. The primary security consideration is that the server's core function is to execute user-defined commands (`cmd` and `args` in `mux.toml`). If a malicious actor gains control over the `mux.toml` configuration file, they could configure `rmcp-mux` to execute arbitrary commands on the system. Therefore, the security relies heavily on the trustworthiness of the configuration source.
Updated: 2025-12-26GitHub→
0
0
Medium Cost
dsactivi-2 icon

lead-ai-mcp

by dsactivi-2

Sec5

This server acts as a central hub for automating lead generation, web scraping for housing and jobs, and orchestrating outbound call campaigns using various AI models for TTS and LLM interactions.

Setup Requirements

  • ⚠️Requires numerous API keys for various external services (e.g., OpenAI, Anthropic Claude, Azure Speech/OpenAI, Google TTS, ElevenLabs, Vonage, Sipgate, Make.com, external Lead Builder, and Cloud Agents backends). These services often incur usage-based costs.
  • ⚠️Requires a Vonage private key file for voice calls (`VONAGE_PRIVATE_KEY` points to a file path) which needs to be securely generated and stored on the server.
  • ⚠️The web scraping functionality relies on Puppeteer, which necessitates a Chromium browser installation (the code defaults to '/snap/bin/chromium' suggesting a Linux environment with Snap).
  • ⚠️Upon first run, if no users are found, a default 'admin' user is created with the password 'admin123'. This password must be changed immediately after setup for security.
Review RequiredView Analysis
CRITICAL: A default 'admin' user is created with password 'admin123' if no users exist. This password must be changed immediately after initial setup to prevent unauthorized access. API keys and secrets are loaded from environment variables, which is good practice, but the system relies on the secure configuration and protection of these keys. The 'ops_pm2_logs' tool uses 'child_process.execAsync' with user-provided input ('process' name), which could pose a command injection risk if not sufficiently sanitized by the underlying 'pm2' CLI. Scraper credentials are encrypted on disk using a key derived from an environment variable ('SCRAPER_SECRET') and the hostname, which is an improvement over plaintext storage but remains a sensitive data point.
Updated: 2026-01-18GitHub→
PreviousPage 547 of 760Next