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
Low Cost
RajTejani61 icon

test-mcp-server

by RajTejani61

Sec9

An application for tracking and managing personal or small business expenses, allowing users to add, list, and summarize financial outlays.

Setup Requirements

  • ⚠️Requires Python 3.13 or newer
Verified SafeView Analysis
The code effectively uses parameterized queries for all SQLite operations, preventing common SQL injection vulnerabilities. There are no apparent uses of 'eval', 'exec', or direct shell commands without proper sanitization. File I/O is restricted to reading local JSON for categories, which is safe. The main security considerations would pertain to the FastMCP framework itself and the file system permissions for the local SQLite database.
Updated: 2025-12-02GitHub→
0
0
Low Cost
Alemusica icon

social-cli-mcp

by Alemusica

Sec9

A CLI tool and MCP server for AI agents to post content across multiple social media platforms (Twitter/X, Reddit, LinkedIn, Instagram).

Setup Requirements

  • ⚠️Requires API credentials for each desired social media platform (Twitter/X, Reddit, LinkedIn, Instagram).
  • ⚠️Instagram posting requires a Business or Creator account linked to a Facebook App.
  • ⚠️Requires specific JSON configuration to integrate with an MCP-compatible AI agent, including the path to the compiled server JS file.
Verified SafeView Analysis
The server uses the Model Context Protocol (MCP) SDK and communicates via standard I/O (stdio), limiting direct network exposure. API keys are loaded from environment variables (.env), preventing hardcoding. Social media interactions are handled by established third-party libraries (twitter-api-v2, snoowrap, custom clients for LinkedIn/Instagram). Input validation schemas are defined for MCP tools. There are no obvious signs of 'eval' usage, code obfuscation, or direct shell command injection points with unsanitized user input. The main security consideration is the secure handling of API credentials by the user.
Updated: 2026-01-18GitHub→
0
0
Low Cost
whobcode icon
Sec7

Deploys a remote Model Context Protocol (MCP) server on Cloudflare Workers, providing simple calculator tools without requiring authentication for public access.

Setup Requirements

  • ⚠️Requires a Cloudflare account and Wrangler CLI for deployment.
  • ⚠️The server is publicly accessible by design (authless), which means anyone can invoke the calculator tools. This is a feature for this demo, but a critical consideration for other use cases.
Verified SafeView Analysis
The server is explicitly designed to be 'authless' as per its name and README, meaning anyone can publicly access and invoke its calculator tools. While the current implementation uses Zod for input validation for simple arithmetic operations, mitigating immediate code injection risks for these specific tools, deploying an authless server with more powerful or sensitive tools would be a significant security vulnerability. No 'eval' or obvious malicious patterns were found in the provided source code. The architecture leverages Cloudflare Durable Objects for the MCP agent, which provides a secure runtime environment.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
ssatama icon
Sec9

Provides an MCP server to enable AI models to search, filter, and retrieve detailed profiles of rescue dogs from European and UK organizations for adoption.

Setup Requirements

  • ⚠️Requires Node.js 18+ runtime
  • ⚠️Requires specific Claude Desktop configuration to register the server
  • ⚠️Relies on the external 'rescuedogs.me' API for all data, so its availability is crucial
Verified SafeView Analysis
The server uses `zod` for robust input validation on all tool parameters, significantly reducing risks from malformed requests. All external API communication is to known and hardcoded `rescuedogs.me` endpoints via `axios`, with comprehensive error handling for network issues, timeouts, and API-specific errors (e.g., 404, 429, 500s). No 'eval' or dynamic code execution is observed. There are no apparent hardcoded secrets or direct file system interactions for sensitive data. An in-memory cache (`node-cache`) is used to improve performance without persisting sensitive data to disk. The project also includes a clear `SECURITY.md` for vulnerability reporting.
Updated: 2026-01-17GitHub→
0
0
Low Cost
diberry icon
Sec9

Generates comprehensive markdown documentation and AI-generated example prompts for Azure MCP server tools.

Setup Requirements

  • ⚠️Requires Docker Desktop or Docker Engine installed and running with at least 8GB RAM and ~10GB free disk space.
  • ⚠️Requires PowerShell 7.4+ to be installed locally.
  • ⚠️Requires Node.js 14.0.0 or higher and npm for CLI output generation.
  • ⚠️Requires .NET 9.0 SDK and a specific .NET 10.0 Preview SDK (e.g., 10.0.100-rc.2.25502.107) for building C# components.
  • ⚠️Azure OpenAI API Key, Endpoint, and Model Name (e.g., FOUNDRY_API_KEY, FOUNDRY_ENDPOINT, FOUNDRY_MODEL_NAME) are required for AI features, which is a paid service.
Verified SafeView Analysis
The system primarily processes local files and interacts with trusted Microsoft services (Azure OpenAI, Azure MCP repository, npm @azure/mcp package). Sensitive API keys are handled via environment variables or .env files, which are correctly masked in logs. Docker containers are configured to run as a non-root user, enhancing security by preventing root-owned files on the host system. No 'eval', obfuscation, or obvious malicious patterns were found. The primary external dependency is the Microsoft/MCP repository cloned into Docker, which is assumed to be trusted.
Updated: 2026-01-17GitHub→
0
0
Low Cost
SahiL911999 icon
Sec10

This repository provides a comprehensive guide for AI engineers to understand and implement Model Context Protocol (MCP) servers for standardized AI system integrations and extending LLM capabilities.

Setup Requirements

  • ⚠️Requires a PDF reader to access the main guide content.
  • ⚠️Assumes proficiency in programming languages (Python, JavaScript/TypeScript) and familiarity with Large Language Models for effective learning.
Verified SafeView Analysis
The provided source code is a documentation repository containing a README and a PDF guide, not an executable server. Therefore, no server-side security vulnerabilities like 'eval', obfuscation, hardcoded secrets, or malicious patterns can be identified or assessed from the given input. The repository itself is static documentation and poses no direct execution risks.
Updated: 2025-12-12GitHub→
0
0
Low Cost
vicharanashala icon

faq-mcp-server

by vicharanashala

Sec6

An intelligent FAQ search and management system leveraging FastMCP, hybrid TF-IDF, and semantic embeddings to provide quick answers and an administrative interface for content updates.

Setup Requirements

  • ⚠️Requires MONGODB_URI environment variable for database connectivity.
  • ⚠️ADMIN_PASSWORD environment variable must be securely set for the Admin server; avoid the insecure 'admin123' default in 'docker-compose-admin.yml'.
  • ⚠️Manual setup requires Python 3.12+ and 'sentence-transformers' (or other embedding library if configured).
Review RequiredView Analysis
The 'docker-compose-admin.yml' file specifies a default 'ADMIN_PASSWORD=admin123', which is a critical security vulnerability if deployed without overriding. While the main 'docker-compose.yml' prompts for it as an environment variable, the presence of this insecure default in the codebase is concerning. Additionally, 'admin_server.py' contains commented-out functions ('last_n', 'download_data') that would write user data to a mounted volume. Although currently disabled, if reactivated or a similar vulnerability exists, it could lead to path traversal or arbitrary file write risks. Authentication for the 'add_faq' tool uses a robust method (secrets.compare_digest for password checking via headers) but relies on correct client-side configuration.
Updated: 2026-01-19GitHub→
0
0
Low Cost
hunter-volkman icon

rovo-mcp-server

by hunter-volkman

Sec1

Facilitating and exploring integrations with Atlassian's Rovo Managed Cloud Platform server.

Review RequiredView Analysis
No source code beyond the README was provided for analysis. Therefore, a comprehensive security audit for 'eval', obfuscation, network risks, hardcoded secrets, or malicious patterns could not be performed. The safety cannot be verified.
Updated: 2025-12-12GitHub→
0
0
Low Cost
Saroj-Shandiliya icon

Vulnerable-MCP_Server

by Saroj-Shandiliya

Sec1

Educational tool for demonstrating and testing Model Context Protocol (MCP) server vulnerabilities.

Review RequiredView Analysis
This server is intentionally designed to be highly vulnerable. It contains critical SQL Injection (direct input concatenation into SQLite queries in `get_user_details`), Command Injection/Remote Code Execution (direct input passed to `child_process.exec` in `cleanup_logs`), a hidden/shadow tool (`debug_access`), simulated data exfiltration (`submit_feedback`), unauthenticated state modification/tool poisoning (`configure_server`), and a CORS misconfiguration (`Access-Control-Allow-Origin: *`). Running this in any non-isolated, non-educational environment is extremely dangerous.
Updated: 2025-12-02GitHub→
0
0
Low Cost
4eyedengineer icon

mcp-server-deployments

by 4eyedengineer

Sec9

Manages Kubernetes deployments for multiple MCP servers using a GitOps approach with ArgoCD.

Setup Requirements

  • ⚠️Requires a Kubernetes cluster (v1.25+) with nginx-ingress controller, cert-manager, and metrics-server pre-installed.
  • ⚠️Requires ArgoCD to be installed and configured in the Kubernetes cluster.
  • ⚠️For private Git repositories, ArgoCD must be configured with repository access credentials (username/PAT).
Verified SafeView Analysis
The repository utilizes robust security practices including strict NetworkPolicies (blocking internal network access, limiting ingress to ingress controller, controlled egress for DNS/HTTPS), ResourceQuotas, LimitRanges, and Kubernetes Pod Security Standards (restricted). No hardcoded secrets or 'eval' patterns are present. The GitOps model itself requires a secure Git repository, as compromise of the repository could lead to cluster compromise. Overall, the configuration itself demonstrates a high level of security awareness.
Updated: 2025-12-02GitHub→
0
0
High Cost
nmeldre icon

ATC_SNOMED_MCP

by nmeldre

Sec5

This server maps medicinal product substance names to SNOMED CT Concept IDs and ATC codes using XML input and web scraping, primarily for Norwegian healthcare data.

Setup Requirements

  • ⚠️Requires outbound internet access to 'dailybuild.terminologi.helsedirektoratet.no' (SNOMED CT API) and 'felleskatalogen.no' (ATC Code web scraping).
  • ⚠️Relies on 'dailybuild.terminologi.helsedirektoratet.no' for SNOMED CT, which implies it might be a development/unstable endpoint.
  • ⚠️Designed for FastMCP Cloud deployment, requiring a FastMCP account and platform understanding for setup and operation.
Review RequiredView Analysis
The server parses user-supplied XML content using `xml.etree.ElementTree.fromstring` without explicitly disabling DTD processing or external entity loading. This may create a potential XML External Entity (XXE) injection vulnerability if malicious XML input is provided. Additionally, the ATC code lookup performs web scraping, which introduces a dependency on external website structure and availability, though it doesn't pose a direct security risk to the server itself. File creation (`generate_output_filename`) is session-based and writes to an 'Output' directory, which is acceptable in this context but not robust against arbitrary path traversal if `input_filename` were directly user-controlled.
Updated: 2025-12-14GitHub→
0
0
Low Cost
kral2 icon
Sec9

This project demonstrates AI-assisted infrastructure and secrets management using HashiCorp's Vault and Terraform Model Context Protocol (MCP) servers within VS Code.

Setup Requirements

  • ⚠️Requires VS Code with MCP support
  • ⚠️Docker required for MCP servers
  • ⚠️Vault server (local dev mode is an option) access required
  • ⚠️Terraform Cloud/Enterprise account (optional, can use local Terraform)
  • ⚠️AWS account with credentials for infrastructure deployment
Verified SafeView Analysis
The project is a demo, and security practices are well-communicated. It guides users to use environment variables for sensitive credentials (AWS, Vault, TFE tokens) and advocates for least-privilege Vault policies. There are no 'eval' statements or obfuscation. The hardcoded local IP (127.0.0.1) is for a development Vault server, with clear instructions on how to set it up. It explicitly warns about AWS costs and the need to destroy resources. While a dev Vault token (`root`) is mentioned, it's explicitly for a dev server and requires user setup.
Updated: 2025-11-28GitHub→
PreviousPage 471 of 713Next