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
Medium Cost
jfomhover icon
Sec9

Manages and validates JSON documents against defined schemas, designed for integration with AI assistants via the Model Context Protocol.

Setup Requirements

  • ⚠️Requires Python 3.11+.
  • ⚠️Manual JSON configuration needed for integration with Claude Dev Extension or Claude Desktop App, including absolute paths.
  • ⚠️Uses local file system for storage, which might require specific folder permissions and management.
Verified SafeView Analysis
The server employs robust schema validation, optimistic locking for concurrency control, and safe JSON Pointer operations to prevent path traversal issues. Storage uses atomic file operations (temp file + rename + fsync) for data integrity. No 'eval' or similar dangerous functions were found. The tool primarily uses local file-based storage, reducing external network attack surfaces. Overall, it demonstrates good security practices within its scope.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
scarnyc icon

jira-mcp-server

by scarnyc

Sec8

Integrates with JIRA to provide a portable Model Context Protocol server, enabling AI clients like Claude Code to interact with JIRA operations.

Setup Requirements

  • ⚠️Requires JIRA_URL, JIRA_USERNAME, and JIRA_API_TOKEN (or JIRA_PERSONAL_ACCESS_TOKEN) environment variables for configuration.
  • ⚠️Requires Python 3.10 or higher.
Verified SafeView Analysis
The server employs Pydantic settings for secure environment variable loading, preventing hardcoded secrets. It uses `httpx` for async network requests with SSL verification. A 'read-only' mode and tool filtering are implemented to enhance security by restricting write operations and exposed functionalities. Input sanitization for JQL queries relies on JIRA's API, which is standard for such integrations. No 'eval' or other obvious malicious patterns were found in the provided source code.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
TalatWaheed icon
Sec8

Hosts a custom Minecraft server and deploys an AI-powered bot that interacts with players and the game world.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid)
  • ⚠️Requires Node.js runtime
Verified SafeView Analysis
The server uses environment variables for sensitive API keys, which is good practice. It creates a Minecraft server and a bot, leveraging external libraries like Mineflayer and OpenAI. The primary security considerations revolve around the inherent risks of AI agents (e.g., prompt injection leading to unintended in-game actions) and external API dependencies. No 'eval' or obvious malicious patterns were found in the provided source.
Updated: 2025-12-11GitHub→
0
0
Low Cost
schimmmi icon

fyta-mcp-server

by schimmmi

Sec9

A Model Context Protocol (MCP) Server for FYTA plant sensor data, enabling an AI assistant like Claude to access and intelligently analyze plant health, sensor data, and care history.

Setup Requirements

  • ⚠️Requires a FYTA account with active plant sensors.
  • ⚠️Python 3.10 or higher is required.
  • ⚠️When configuring for Claude Desktop, absolute paths to the Python executable and working directory are critical and commonly lead to setup errors.
Verified SafeView Analysis
The server uses environment variables for sensitive credentials (FYTA_EMAIL, FYTA_PASSWORD) which is a good practice. It explicitly warns against hardcoding secrets and provides guidance for securing exposed HTTP endpoints (e.g., with HTTPS and authentication in production). The core MCP server uses standard I/O and does not open network ports by default. Local file storage for care actions and plant context is used, reducing remote data exposure risks. No 'eval' or obvious malicious patterns were detected in the source code.
Updated: 2026-01-18GitHub→
0
0
High Cost
victorRT3868 icon

shopping-agent-ai

by victorRT3868

Sec4

An AI-powered shopping assistant for cross-platform price comparison and objective buying advice on e-commerce platforms (JD, Taobao, Vipshop).

Setup Requirements

  • ⚠️Requires an LLM API Key (e.g., OpenAI, DeepSeek), which is typically a paid service.
  • ⚠️Manual login/QR code scanning is required for Taobao and Xiaohongshu scrapers due to anti-bot measures.
  • ⚠️The JD OCR mode (`jd_gui.py`) takes over your mouse and keyboard, demanding full user attention and no interference during its operation.
  • ⚠️Initial setup for PaddleOCR might require downloading models, necessitating internet access and available disk space.
Verified SafeView Analysis
1. Browser session cookies are stored in plain JSON files (`auth.json`, `auth_zhihu.json`, `auth_xhs.json`), exposing login data if the host machine is compromised. 2. `src/scrapers/jd_gui.py` utilizes `pyautogui` to directly control the user's mouse and keyboard, which presents a significant operational risk; users must be fully aware and avoid system interaction during its execution. 3. Playwright in `headless=False` mode with anti-detection scripts, though necessary for certain platforms, might still trigger anti-bot measures, potentially leading to temporary IP bans or account restrictions on target e-commerce sites.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
gomcpgo icon
Sec7

An MCP server that enables LLMs to create and edit HTML documents with embedded styles, images, and videos.

Setup Requirements

  • ⚠️Requires Pandoc for PDF/DOCX export (and Texlive for robust PDF support).
  • ⚠️Requires headless Chrome/Chromium for optimal PDF export (go-rod attempts auto-download if not found, which can be a large download).
Verified SafeView Analysis
The server uses `os/exec` to run external tools like `pandoc` and `go-rod` (headless Chrome) for document export. While the arguments are constructed programmatically, relying on external executables always introduces a potential attack surface if those tools have vulnerabilities or if crafted HTML content can exploit them for command injection. No direct `eval` or obvious shell injection points were found within the application's argument handling. `source_path` for `add_media` copies files without executing them, mitigating direct execution risk for media files. No hardcoded secrets or unexpected network activities beyond Chrome auto-download are present.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
titan77champion icon

titan77champion.github.io

by titan77champion

Sec10

Effortlessly extracts clean, LLM-ready markdown data from any website, including Cloudflare-protected sites, for web scraping purposes.

Setup Requirements

  • ⚠️Requires downloading and installing an executable binary from GitHub releases, necessitating trust in the distributed application.
  • ⚠️An internet connection is required for downloading content.
Verified SafeView Analysis
The provided source code is an `index.md` file, which is documentation for a separate desktop application called 'webustler'. This markdown file itself contains no executable code or inherent security vulnerabilities. The security of the 'webustler' application described is not assessed as its actual source code was not provided.
Updated: 2026-01-19GitHub→
0
0
Low Cost
rimaslogic icon

n8n-monitor-docker

by rimaslogic

Sec8

Facilitates AI (Claude) monitoring and management of n8n workflows through a Docker-based server.

Setup Requirements

  • ⚠️Requires an existing Docker environment on the deployment server.
  • ⚠️Requires SSH access to the deployment server, with the specified user having permissions to execute docker commands.
  • ⚠️Requires an active n8n instance and a specific n8n webhook URL to be configured.
  • ⚠️Requires manual download and import of a pre-built n8n workflow JSON file.
Verified SafeView Analysis
The provided source code (docker-compose.yml, README.md, deploy.sh) does not contain obvious vulnerabilities such as 'eval', code obfuscation, or hardcoded secrets. It leverages standard Docker and SSH for communication. The security of the overall setup heavily relies on the underlying SSH configuration of the host server and Docker environment security. The client-side configuration's use of 'StrictHostKeyChecking=accept-new' for SSH is a convenience for initial setup and might be considered less strict than explicit host key verification for production environments.
Updated: 2025-12-14GitHub→
0
0
Low Cost
rioadel icon

MCP_Servers

by rioadel

Sec4

Building and deploying Model Context Protocol (MCP) servers using C# and .NET for integration with AI assistants like Copilot Chat.

Setup Requirements

  • ⚠️Requires .NET SDK to build the server project.
  • ⚠️Requires a Google Generative AI API Key (potentially a paid service).
  • ⚠️Requires specific IDE configuration (.mcp.json) for VS Code or Visual Studio to integrate with AI assistants.
Review RequiredView Analysis
The `appsettings.json` file contains a hardcoded Google Generative AI API key. While a comment suggests using user secrets in production, its presence in a source-controlled file is a critical security vulnerability as it could expose credentials if the repository is public or handled improperly.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
martinremy icon

orgmode-mcp

by martinremy

Sec9

An MCP server enabling AI clients like Claude Desktop to interact with Org Mode files for task management, note-taking, and structured data workflows.

Setup Requirements

  • ⚠️Requires Node.js 18 or later to run.
  • ⚠️Manual configuration of `config.json` is required to specify Org Mode file paths (supports glob patterns and tilde expansion). Incorrect paths will result in errors or no files being found.
  • ⚠️For integration with Claude Desktop, the absolute path to the server's `dist/index.js` and `config.json` must be manually configured in Claude Desktop's `claude_desktop_config.json`.
Verified SafeView Analysis
The server communicates primarily via standard I/O (stdio) using the Model Context Protocol (MCP), which means it typically runs as a local process and does not expose network ports, significantly reducing the external attack surface. It uses Zod for input validation for tool and prompt arguments, limiting the scope of client-controlled input. The core functionality involves reading local Org Mode files specified in a user-provided `config.json`. The primary security consideration is the user's responsibility in configuring the `orgFiles` array in `config.json`. If sensitive paths are included, the server will expose their content to the connected MCP client. However, the server itself does not perform arbitrary file writes, code execution from file content, or arbitrary path traversal outside of the explicitly configured glob patterns. The file parsing and content retrieval logic (`orgParser.ts`) appears to handle file reading safely without dynamic code execution.
Updated: 2026-01-14GitHub→
0
0
Medium Cost
ericmjl icon
Sec3

A personal productivity MCP server designed to assist developers with tasks such as task prioritization, work logging, Git workflows, code reviews, debugging, and Obsidian knowledge management.

Setup Requirements

  • ⚠️Requires `pixi` for dependency management and command execution.
  • ⚠️Relies heavily on `gh cli` for GitHub-related tasks (e.g., debugging, dev digest, code review).
  • ⚠️Requires a local Obsidian vault for knowledge management prompts (MOCs).
  • ⚠️Specifically pins `fastmcp==2.0.0` and `pydantic>=2.0.0,<3.0.0` due to compatibility issues with newer versions.
  • ⚠️HTTP transport is not yet implemented, only stdio transport is supported.
Review RequiredView Analysis
The server uses prompts that instruct the LLM to generate and execute shell commands (e.g., `git checkout -b <branch-name>`) and perform file system operations (e.g., creating/updating Obsidian notes based on `topic` or `note_name`). There is no explicit sanitization of LLM-generated strings before they are used in these contexts. This creates a significant risk of command injection (if a malicious branch name is generated) or path traversal (if a malicious note name/topic is generated), potentially allowing the LLM to execute arbitrary commands or write files to unintended locations. Additionally, the `upgrade_repo_to_template` prompt instructs cloning external repositories, which carries a supply chain risk, even though the current URL is benign. While the LLM is instructed with rules, these are not programmatic safeguards.
Updated: 2025-12-17GitHub→
0
0
Medium Cost
junzzhu icon
Sec6

A Model Context Protocol (MCP) server for OpenShift diagnostics and troubleshooting.

Setup Requirements

  • ⚠️Requires 'oc' command line tool installed and in PATH.
  • ⚠️Requires 'oc' to be authenticated to the target OpenShift cluster.
  • ⚠️GPU monitoring tools require NVIDIA GPU Operator and DCGM exporter installed.
  • ⚠️Monitoring tools rely on OpenShift Monitoring (Prometheus Metrics API) being available.
  • ⚠️Requires Python 3.10 or higher.
Verified SafeView Analysis
The server executes `oc` commands using `asyncio.create_subprocess_exec`, which is generally safe as arguments are passed as a list, mitigating shell injection risks. However, the `_query_via_route` function uses `curl -k` (insecure) to query Prometheus, disabling TLS certificate validation. This is a security vulnerability that could allow man-in-the-middle attacks if the network path to Prometheus is not fully trusted. Additionally, the `run_oc_debug_node` function allows running arbitrary shell scripts (`bash -c '...'`) on a node, a powerful primitive. While currently only used with hardcoded scripts within `inspect_node_storage_forensics` (making it safe in this context), this function could be a high-risk vector if ever exposed to unsanitized user input.
Updated: 2025-12-15GitHub→
PreviousPage 553 of 713Next