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
federicodeponte icon

openclaude

by federicodeponte

Sec9

A comprehensive configuration setup for the Claude Code IDE, enabling remote VM development, file synchronization, SSH tunneling, and integration with 14 specialized Model Context Protocol (MCP) servers for enhanced AI-driven development.

Setup Requirements

  • ⚠️Requires Node.js and npm for installing MCP servers.
  • ⚠️Requires API keys for multiple external services (e.g., Context7, Snyk, GitHub, CodeRabbit, Brave Search, Supabase), some of which may be paid.
  • ⚠️Remote VM development requires a configured VM with SSH access and proper SSH key setup.
  • ⚠️This configuration is designed for the Claude Code IDE, implying its prior installation.
Verified SafeView Analysis
The repository demonstrates a strong focus on security, providing explicit guidelines for credential management (e.g., 'NEVER Commit These Patterns' and pre-commit checks), recommending SSH key authentication, firewalls, VPNs, and regular updates. It includes commands for comprehensive security analysis using Semgrep. Placeholders for API keys and sensitive paths are intentionally left for the user to configure, reducing the risk of hardcoded secrets in the distributed code. The main security risks stem from potential user misconfiguration (e.g., not replacing placeholders, using weak SSH keys) rather than inherent vulnerabilities in the provided scripts. No 'eval' or malicious obfuscation was found.
Updated: 2025-12-18GitHub→
0
0
High Cost
muskke icon
Sec4

Remotely control Adobe Illustrator to automate design tasks, generate graphics via scripts, and visualize results with screenshots.

Setup Requirements

  • ⚠️Requires Adobe Illustrator (2022+ recommended) to be installed and running.
  • ⚠️On macOS, the terminal/client running the MCP server needs screen recording permissions for the `view` tool to capture screenshots.
  • ⚠️On macOS, the `view` tool captures a fixed screen region (0,0,960,1080), which may not perfectly align with the Illustrator window or desired content on all screen setups.
Review RequiredView Analysis
The `run` tool executes arbitrary ExtendScript/JavaScript code directly within Adobe Illustrator via COM (Windows) or AppleScript (macOS). While this is the core intended functionality, it means any script provided by the MCP client is executed with the privileges of the Illustrator application. This poses a significant security risk if the client or the source of the scripts is untrusted, as malicious scripts could potentially interact with the file system, network, or other system components via Illustrator's scripting environment. No explicit Python `eval` or hardcoded secrets are found in the server's Python code itself. Robust error handling is present, converting exceptions into `TextContent` responses instead of crashing the server.
Updated: 2025-11-30GitHub→
0
0
High Cost
ProdMoh icon

prodmoh-mcp

by ProdMoh

Sec9

Connects product requirements (PRDs, user stories) from ProdMoh to AI clients and IDEs for context-aware code generation, testing, and clarification.

Setup Requirements

  • ⚠️Requires a ProdMoh Dashboard account to generate an 'x-prodmoh-token'.
  • ⚠️Requires manual configuration of 'mcp.json' in a compatible AI client (e.g., Cursor, VS Code).
Verified SafeView Analysis
The provided source code is primarily client-side configuration (mcp.json) and documentation for connecting to a hosted ProdMoh MCP service via Server-Sent Events (SSE) over HTTPS. It does not contain server-side code to audit for 'eval', obfuscation, or network risks. The client configuration requires a user-generated token (x-prodmoh-token) for authentication, which is a secure practice. No malicious patterns were detected in the provided client configuration.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
thekaranpargaie icon

kube-mcp

by thekaranpargaie

Sec2

Facilitates Kubernetes cluster management and operations through natural language commands, acting as an AI assistant backend for developer and DevOps workflows.

Setup Requirements

  • ⚠️Requires Node.js v18+ to be installed on the host system.
  • ⚠️Requires `kubectl` to be installed and configured with appropriate access to a Kubernetes cluster.
  • ⚠️Requires `jq` to be installed on the host system for the `get-pod-metrics` tool to function correctly, although this is not explicitly listed in requirements.
  • ⚠️The server communicates via `stdio` transport, necessitating an MCP-compatible client (e.g., Claude Desktop, Cursor AI, VS Code MCP extension) configured to launch it as a child process.
Review RequiredView Analysis
The server directly constructs and executes `kubectl` commands using `child_process.exec` where many arguments (e.g., pod names, commands for `exec`, file paths for `apply` and `cp`, patch data for `patch`, labels/annotations) are derived directly from user input without apparent sanitization. This is a severe command injection vulnerability, allowing a malicious actor to potentially execute arbitrary code on the host system where the MCP server is running. For example, injecting shell metacharacters into the `command` argument of the `exec` tool, or the `file` argument of `apply`, could lead to Remote Code Execution. The `port-forward` tool also has the potential to expose internal services if misused. Running this server requires significant trust in the input it receives and robust sanitization of all user-controlled arguments is critical.
Updated: 2025-11-28GitHub→
0
0
High Cost
hjkim7796 icon
Sec7

Serves a Retrieval-Augmented Generation (RAG) system via HTTP/SSE using the Model Context Protocol (MCP), primarily for integration with AI assistants like Claude Desktop.

Setup Requirements

  • ⚠️Requires `ANTHROPIC_API_KEY` environment variable for the 'rag_query' tool, which utilizes a paid LLM API (Claude).
  • ⚠️For enhanced PDF processing including OCR, external Tesseract-OCR software and associated Python libraries (`pytesseract`, `pdf2image`, `Pillow`) must be installed.
  • ⚠️Utilizes a persistent local ChromaDB, with data stored in the `./chroma_db` directory.
  • ⚠️The server defaults to running on port 8000, which may conflict with other applications.
Verified SafeView Analysis
The default configuration uses broad CORS (`allow_origins=['*']`) and does not enforce API key authentication by default, making it insecure for public production deployment. The `ANTHROPIC_API_KEY` is loaded from environment variables, preventing hardcoding of LLM secrets. The README explicitly advises adding API key authentication, HTTPS, and rate limiting for production environments. No 'eval' or other directly malicious patterns were found in the provided code.
Updated: 2025-12-04GitHub→
0
0
Low Cost
Hiruna219066 icon

poly-cloud-mcp

by Hiruna219066

Sec9

Unified Model Context Protocol (MCP) server for managing resources across AWS, Google Cloud, Azure, and DigitalOcean via their respective CLI tools.

Setup Requirements

  • ⚠️Requires Deno runtime to be installed.
  • ⚠️Requires respective cloud provider CLI tools (aws, gcloud, az, doctl) to be installed and configured locally.
  • ⚠️Requires Deno to be run with `--allow-run`, `--allow-read`, `--allow-env`, `--allow-write`, and `--allow-net` permissions.
Verified SafeView Analysis
The server's core functionality involves executing external CLI commands (`aws`, `gcloud`, `az`, `doctl`) using Deno's `--allow-run` permission, which is explicitly noted and justified. Command arguments are constructed carefully from JSON inputs, minimizing direct shell injection risks through the server itself. It leverages standard CLI authentication methods (e.g., local config files, environment variables like AWS_*, DIGITALOCEAN_ACCESS_TOKEN, or login sessions), avoiding hardcoded secrets. The server operates in a local-agent mode using stdio, meaning it doesn't expose a network attack surface for its own communication, though the wrapped CLIs will make network calls. The project includes a SECURITY.md and uses GitHub's security features, indicating a strong security posture.
Updated: 2026-01-19GitHub→
0
0
Low Cost
Temple-of-Epiphany icon

mcp-web-pro

by Temple-of-Epiphany

Sec4

A Model Context Protocol (MCP) server for Claude Desktop that provides local web content preview (HTML, JSX, images) generated by Claude.

Setup Requirements

  • ⚠️Requires Python 3.10+.
  • ⚠️Requires Claude Desktop (latest version) for integration.
  • ⚠️Configuration requires absolute paths for the server command and preview root in Claude Desktop's config file.
  • ⚠️JSX transpilation relies on PyExecJS and requires Node.js to be installed and available on the system, which is a non-obvious dependency for a Python project.
Review RequiredView Analysis
The server is designed for localhost-only access, which mitigates external network risks. Path validation is implemented to prevent directory traversal. However, the documentation explicitly states that HTML and JSX content is *not* sanitized by default. The `render_jsx` tool uses PyExecJS, which can execute arbitrary JavaScript. This creates a high risk of Cross-Site Scripting (XSS) and potential arbitrary code execution within the user's local environment if Claude generates or processes untrusted HTML/JSX content without proper sanitization. Users are warned to only render trusted content. The project incorporates Snyk for security scanning in its CI/CD.
Updated: 2025-11-27GitHub→
0
0
Low Cost
alexandremattioli icon

CloudStack-MCP-Server

by alexandremattioli

Sec1

Unable to determine without source code. The project name 'CloudStack-MCP-Server' suggests a server component potentially related to a CloudStack Multi-Cloud Platform or Management Control Plane.

Setup Requirements

  • ⚠️Source code was not provided, so specific setup requirements (e.g., Docker, specific Python versions, API keys) could not be identified.
  • ⚠️The server's functionality and dependencies are unknown without code.
Review RequiredView Analysis
CRITICAL: No source code was provided for analysis. A comprehensive security audit could not be performed, and the presence of dangerous patterns (like 'eval', hardcoded secrets, or malicious code) cannot be verified. Running this server without source code review is extremely risky.
Updated: 2025-11-20GitHub→
0
0
Medium Cost
WycliffeAssociates icon

DataverseMetadataMCPServer

by WycliffeAssociates

Sec8

Provides AI assistants with tools to query Microsoft Dataverse/Power Platform metadata.

Setup Requirements

  • ⚠️Requires .NET 8.0 SDK for building from source.
  • ⚠️Requires a Dataverse environment with appropriate access credentials.
  • ⚠️Requires a correctly formatted Dataverse connection string to be set as the `DATAVERSE_CONNECTION_STRING` environment variable.
Verified SafeView Analysis
The server relies on the `DATAVERSE_CONNECTION_STRING` environment variable for authentication, which is a good practice for handling secrets. It uses standard Microsoft libraries for Dataverse interaction and the Model Context Protocol SDK for server communication via standard I/O (stdio), limiting direct network exposure. Without the full C# source code, a thorough check for dynamic code execution (e.g., `eval` equivalents) or other low-level vulnerabilities cannot be performed. However, the stated purpose (metadata retrieval) and the use of well-known libraries suggest a generally secure design.
Updated: 2025-12-11GitHub→
0
0
Medium Cost
thedevopsstore icon

email-mcp-server

by thedevopsstore

Sec9

Provides Model Context Protocol (MCP) interface for Microsoft 365 Outlook email operations, primarily for autonomous agents.

Setup Requirements

  • ⚠️Requires Azure App Registration with specific Application permissions (Mail.Read, Mail.ReadWrite, Mail.Send) and administrator consent.
  • ⚠️For shared mailboxes, the MS365_USER_IDENTIFIER environment variable or a specific tool parameter is mandatory when using Client Credentials Flow.
  • ⚠️When deploying to AWS AgentCore Runtime, the MCP server *must* internally listen on '0.0.0.0:8000' (not 8100 or localhost), as specified in the deployment guide.
Verified SafeView Analysis
The server design emphasizes security: client secrets are expected via environment variables (not hardcoded), and a 'token-only' mode (server_token_only.py) is available that avoids handling client secrets altogether, relying on agents to provide pre-acquired Microsoft Graph access tokens via custom HTTP headers. This 'token-only' mode also explicitly avoids caching client instances by token, preventing long-lived tokens in memory. All network communication to Microsoft Graph uses HTTPS. It listens on '0.0.0.0' for containerized deployments, which is standard but requires network-level security. No 'eval' or other directly dangerous functions were found. Proper Azure AD app registration and permission configuration, along with secure secret management (as advised in documentation), are crucial for its secure operation.
Updated: 2026-01-07GitHub→
0
0
Low Cost
JS2330 icon
Sec10

This repository provides a placeholder or an empty starting point for an MCP server, likely intended for demonstrating a basic server structure, but no functional source code was provided for analysis.

Verified SafeView Analysis
No source code was provided for analysis. Therefore, no code-related security risks, such as 'eval', obfuscation, network risks, hardcoded secrets, or malicious patterns, could be identified. The server itself has no executable code.
Updated: 2025-12-01GitHub→
0
0
Medium Cost
GlacierEQ icon
Sec3

A comprehensive backend server boilerplate for developing decentralized applications with smart contract integration, secure sandboxed execution, and advanced data handling capabilities like vector operations.

Setup Requirements

  • ⚠️Requires a blockchain client/environment (e.g., Ganache, Hardhat) for smart contract interaction
  • ⚠️Requires E2B sandboxed execution environment setup
  • ⚠️Requires a vector database (e.g., Pinecone, Weaviate, Chroma) or an equivalent in-memory solution for vector operations
Review RequiredView Analysis
Source code provided is limited to README.md. Cannot perform a meaningful security audit for 'eval', obfuscation, network risks, hardcoded secrets, or malicious patterns without access to the actual code files. Running without proper code review is not recommended.
Updated: 2025-11-23GitHub→
PreviousPage 638 of 713Next