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)

30
1
Low Cost
Chisanan232 icon

slack-mcp-server

by Chisanan232

Sec9

Provides a robust Model Context Protocol (MCP) server for Slack integration, offering standardized access to Slack's API features through MCP tools and real-time webhook event processing for AI assistants and automation.

Setup Requirements

  • ⚠️Requires Python 3.12+ for optimal performance and modern language features.
  • ⚠️Requires a valid Slack Bot Token (`SLACK_BOT_TOKEN`) for Slack API interactions. A Slack Signing Secret (`SLACK_SIGNING_SECRET`) is mandatory for webhook verification.
  • ⚠️Using Redis or Kafka as a message queue backend (`QUEUE_BACKEND`) requires external Redis or Kafka instances to be set up and configured via `REDIS_URL` or `KAFKA_BOOTSTRAP`.
Verified SafeView Analysis
The project uses Slack's `SignatureVerifier` with `SLACK_SIGNING_SECRET` for webhook request validation, which is a strong security practice. Environment variables are used for sensitive information like Slack tokens, with clear handling via `os.getenv` and `.env` file loading. No instances of `eval`, `exec`, or `subprocess.run(shell=True)` were found. The server can bind to `0.0.0.0` by default for HTTP transports, which is common in Docker but should be secured with appropriate network configurations (e.g., firewall rules, restricted CORS origins) in production environments.
Updated: 2026-01-14GitHub→
30
1
Medium Cost
fabric-testbed icon

fabric_mcp

by fabric-testbed

Sec9

Provides a Model Context Protocol (MCP) server to expose FABRIC Testbed API and inventory queries to LLM clients.

Setup Requirements

  • ⚠️Requires a valid FABRIC ID Token (obtained from FABRIC Portal) for all authenticated calls.
  • ⚠️Requires user-provided SSL certificates (fullchain.pem, privkey.pem) for the NGINX reverse proxy for secure HTTPS access.
  • ⚠️Docker and Docker Compose are the recommended deployment method, though a local Python run is possible.
Verified SafeView Analysis
The server uses token-based authentication (Bearer FABRIC ID token) passed via HTTP headers and explicitly states it does not store user credentials. Sensitive configurations like API hosts are loaded from environment variables. Logging explicitly redacts tokens and recommends terminating TLS at the NGINX reverse proxy. No 'eval' or obvious malicious patterns were found. Relies on external SSL certificate provisioning for NGINX.
Updated: 2025-12-11GitHub→
30
1
Low Cost
GuillaumeMilan icon

mcp_server

by GuillaumeMilan

Sec9

Provides an Elixir DSL for defining and serving Model Context Protocol (MCP) tools, prompts, and resources to AI agents, enabling them to interact with custom backend logic.

Setup Requirements

  • ⚠️Requires a working Elixir and Erlang/OTP environment to develop and run applications using this library.
  • ⚠️Relies on a Plug-compatible HTTP server like Bandit (`{:bandit, "~> 1.0"}`) for serving the MCP endpoints, which must be integrated into the application's supervision tree.
  • ⚠️Developers migrating from versions prior to v0.4.0 must adapt to significant breaking changes, including adding a 'conn' parameter to all controller functions, using struct notation (`.field`) instead of map notation (`["field"]`) for data access, and handling tuple returns from router list functions.
Verified SafeView Analysis
The library itself (McpServer) demonstrates a strong focus on type safety and input validation through its struct-based API and schema definitions, which inherently reduces common vulnerabilities. No 'eval', obfuscation, hardcoded secrets, or obvious malicious patterns were found in the provided source code snippets. The documentation explicitly advises respecting 'recommended security options for MCP servers' for the HTTP transport layer, indicating that deployment-level security (e.g., HTTPS, authentication, access control) is the responsibility of the application developer. Without access to the full `McpServer.HttpPlug` implementation or a complete application's configuration, a comprehensive security audit of a live deployment is not possible, but the library's design principles are sound.
Updated: 2025-12-11GitHub→
30
1
High Cost
1999AZZAR icon
Sec6

Orchestrates intelligent tool chaining, route optimization, and advanced thinking for complex development workflows and enterprise capabilities across an MCP ecosystem.

Setup Requirements

  • ⚠️Requires 'GITHUB_TOKEN' environment variable (a paid GitHub Personal Access Token) for 'awesome-copilot' integration, which is a key feature.
  • ⚠️Heavily relies on discovering and correctly configuring external MCP servers, including a specific hardcoded path for the 'awesome-copilot' dotnet executable (e.g., '/tmp/awesome-copilot-mcp/...'), making setup complex.
  • ⚠️Requires Node.js version 18.0.0 or higher.
Review RequiredView Analysis
The server's core functionality involves discovering and spawning external Model Context Protocol (MCP) servers (e.g., 'awesome-copilot', 'sequential-thinking') and executing their tools. This introduces an inherent security risk, as the server's security is dependent on the trustworthiness and security of all discovered and executed external processes. Sensitive information like 'GITHUB_TOKEN' is passed to these spawned processes via environment variables. The 'MEMORY_FILE_PATH' can be configured via an environment variable, which could pose a risk if an attacker can control this path to a sensitive location. While the server itself avoids direct dangerous patterns like 'eval' on untrusted input, its role as an orchestrator means it's a central point for potential vulnerabilities arising from the ecosystem it manages.
Updated: 2026-01-10GitHub→
30
1
Low Cost
genesisdb-io icon
Sec3

Provides AI clients access to GenesisDB events, enabling natural language queries against the database.

Setup Requirements

  • ⚠️Requires an existing GenesisDB instance to connect to.
  • ⚠️Requires downloading a specific binary for your OS and architecture.
  • ⚠️The primary integration example is with Claude Desktop, requiring specific configuration within its settings.
  • ⚠️Requires a GenesisDB authentication token.
Review RequiredView Analysis
The server is distributed as a pre-compiled binary, and its source code is not provided for security auditing. As a 'Preview Release', it may contain unaddressed vulnerabilities. Running untrusted binaries carries inherent risks. The configuration requires providing a sensitive `GENESISDB_AUTH_TOKEN`.
Updated: 2025-12-01GitHub→
30
1
Low Cost
chirag127 icon
Sec9

Provides a Model Context Protocol (MCP) API for AI agents to retrieve precise, locale-aware date and time data across timezones.

Setup Requirements

  • ⚠️Requires Node.js (version 18 or higher as indicated by `package.json` and `tsconfig.json`).
  • ⚠️Designed to run within a Model Context Protocol (MCP) environment and communicates via standard I/O (stdio), not as a standalone HTTP server by default.
Verified SafeView Analysis
The server uses `zod` for input validation for tool parameters and includes explicit timezone validation. There are no obvious `eval` statements, direct shell command executions from unvalidated user input, or hardcoded sensitive credentials within the provided source code. Communication over standard I/O (stdio) as an MCP service limits direct network attack surfaces, relying on the security of the encompassing MCP runtime.
Updated: 2026-01-18GitHub→
30
1
Medium Cost
JM-Addington icon

mcp-autotask-search

by JM-Addington

Sec9

Exposes Autotask search functionality for tickets, companies, and contacts to LLMs like Claude via the Model Context Protocol.

Setup Requirements

  • ⚠️Requires a separate Autotask Django API server to be running and accessible (default: http://localhost:8000).
  • ⚠️Requires Python 3.10 or higher.
  • ⚠️A valid `AUTOTASK_API_KEY` is mandatory for authentication with the Autotask API.
  • ⚠️When configuring for Claude Desktop/Cursor, the `command` path to `run.sh` must be absolute.
Verified SafeView Analysis
The server uses `httpx` for API calls and retrieves the `AUTOTASK_API_KEY` from environment variables, which is good practice. The `BASE_URL` defaults to `http://localhost:8000`, limiting direct external exposure unless explicitly configured. There are no `eval` calls or clear obfuscation. Potential risks include the security of the underlying Autotask Django API server and ensuring the `AUTOTASK_API_KEY` itself is securely managed and has appropriate permissions.
Updated: 2026-01-14GitHub→
30
1
Medium Cost
Stig-Johnny icon

claude-memory-mcp

by Stig-Johnny

Sec8

Provides persistent memory capabilities for Claude Code agents, allowing them to store and recall decisions, error solutions, project context, learnings, and session state across conversations, with optional cloud synchronization.

Setup Requirements

  • ⚠️Requires manual installation steps including cloning the repository, installing Node.js dependencies (with an optional Firestore package), copying slash command files, and manually configuring ~/.claude/settings.json.
  • ⚠️Optional Google Cloud Firestore sync setup is multi-step, requiring creating a Firebase project, setting up a Firestore database, generating a service account key, and configuring a local memory-config.json file.
  • ⚠️The `@google-cloud/firestore` package must be installed separately if cloud sync is desired.
Verified SafeView Analysis
The server uses SQLite locally and prepared statements, mitigating common SQL injection risks. Optional cloud sync with Google Cloud Firestore requires users to generate and secure a service account key (`firestore-key.json`), which should never be committed to Git. Users are explicitly warned against storing secrets (e.g., API keys, passwords) directly in memory, recommending environment variables instead. Firestore security rules are the user's responsibility for production use. Overall, the code itself is not inherently insecure, but relies on proper user configuration and handling of external secrets and cloud service security.
Updated: 2026-01-18GitHub→
30
1
Low Cost
KWR-Water icon

epanet-mcp-server

by KWR-Water

Sec3

Provides tools for fetching real-time weather information from the NWS API and running EPANET water network simulations.

Setup Requirements

  • ⚠️The `epyt` library, used for EPANET simulations, is a required dependency but is not listed in `pyproject.toml`'s main `dependencies` array and needs to be installed separately.
  • ⚠️The `get_inp_files` tool uses a hardcoded absolute path (`C:/Users/.../models`) for the models directory. This path must be manually changed or exactly match the environment for the server to function correctly.
  • ⚠️The server requires Python 3.11 or newer.
Review RequiredView Analysis
The server has a critical path traversal vulnerability in the `run_epanet_simulation` tool. A manipulated `file_name` argument (e.g., `../secrets.txt`) could allow the server to access and potentially process arbitrary files outside the intended `models` directory. Additionally, the `get_inp_files` tool uses a hardcoded absolute path (`C:/Users/.../models`) which is a security risk (exposing internal system structure) and severely impacts portability. No hardcoded secrets were found in the API keys or critical credentials, but the file system access issues are significant.
Updated: 2026-01-12GitHub→
30
1
Medium Cost
menscheck icon

fourtrader-mcp

by menscheck

Sec3

Serves as a trading engine for Four.meme tokens on BNB Smart Chain, providing AI-driven insights and executing trades.

Setup Requirements

  • ⚠️Requires Wallet Private Key for trading functions, which must be provided in the .env file.
  • ⚠️Requires a Bitquery API Key for comprehensive market data; many data-fetching tools will return limited information without it.
  • ⚠️Designed for integration with Claude Desktop, requiring manual modification of 'claude_desktop_config.json' with the full path to the project.
Review RequiredView Analysis
The trading functions ('buy_token' and 'sell_token') have a critical security vulnerability: 'minTokensOut' and 'minBnbOut' are hardcoded to 0. This means transactions are executed without any slippage protection, exposing user funds to potential 100% loss due to price fluctuations or malicious actors. Additionally, 'poolAddress' in 'buy_token' is hardcoded, which might not be universally correct or robust. While private keys and API keys are correctly managed via environment variables and security documentation is present, the fundamental lack of slippage control makes the trading aspect highly unsafe.
Updated: 2026-01-19GitHub→
30
1
Medium Cost
fabriziosalmi icon

nexus-mcp-server

by fabriziosalmi

Sec7

The Nexus MCP Server provides a micro-capability platform (MCP) exposing a wide array of developer tools via a REST API, enabling tasks from mathematical operations and cryptographic functions to system management, network diagnostics, cloud resource optimization, and dynamic code execution across multiple programming languages.

Setup Requirements

  • ⚠️Requires several Python libraries (e.g., `psutil`, `cryptography`, `httpx`, `dnspython`, `PyPDF2`, `pdfplumber`) to be installed.
  • ⚠️Relies on external system binaries (`git`, `ping`, `traceroute`, `whois`, `systemctl`, `nmap`, `curl`, `wget`, `node`) which must be available on the host or in the Docker image.
  • ⚠️Optimal and secure execution of dynamic code (Python/JavaScript) is designed to run within a Dockerized sandbox environment, necessitating Docker daemon availability.
Review RequiredView Analysis
The server includes extensive capabilities for dynamic code execution (Python, JavaScript) and shell command execution, as well as system/process management and network scanning. Developers have implemented significant security measures including input validation, whitelisting/blacklisting for dangerous operations/imports, resource limits for code execution, and filesystem sandboxing for file operations. However, executing arbitrary user-provided code/commands, even with safeguards, inherently carries risks of bypasses or resource exhaustion. Reliance on external system binaries also introduces a dependency on their security. Running this server in a tightly controlled Docker environment with minimal privileges, as suggested by the `docker-compose.yml`, is crucial to mitigate these inherent risks.
Updated: 2025-12-03GitHub→
30
1
Medium Cost
pollychrome icon

joan-mcp

by pollychrome

Sec9

Enables AI assistants to interact with and manage projects, tasks, goals, milestones, and notes within the Joan productivity app.

Setup Requirements

  • ⚠️Requires interactive browser login for authentication (OAuth flow).
  • ⚠️Requires Node.js version 18 or higher.
  • ⚠️Claude Code restart is necessary after initial setup for changes to take effect.
Verified SafeView Analysis
The server uses strong cryptographic methods (AES-256-GCM, scryptSync) for local authentication token storage and restricts file permissions (0o600). OAuth is used for authentication via a local HTTP callback server, which is well-controlled and times out. Usage of `child_process.spawn` for opening the browser and configuring Claude CLI is for hardcoded commands or locally-derived URLs, reducing the risk of arbitrary command injection. JWT parsing is used only for expiration checks and email extraction, not for authentication verification, which relies on the backend API call. No direct `eval` or blatant obfuscation detected.
Updated: 2026-01-17GitHub→
PreviousPage 208 of 713Next