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
mlgill icon
Sec9

Provides an MCP server to send Pushover notifications from MCP-compatible tools like Cursor, offering customizable alerts.

Setup Requirements

  • ⚠️Requires a Pushover account with a registered application (App Token) and User Key.
  • ⚠️Requires Python 3.10+ and the `uv` package manager installed for direct Python execution.
  • ⚠️When using Docker Compose, requires a `.env` file with `PUSHOVER_TOKEN` and `PUSHOVER_USER_KEY`.
Verified SafeView Analysis
The server correctly loads API credentials from environment variables or a configuration file, avoiding hardcoded secrets. It uses `httpx` for network requests and `FastMCP` for server functionality, both reputable libraries. Input validation (e.g., message length, priority range) is performed. When run in SSE (web service) mode, standard web service security practices (e.g., firewall, access control) should be considered if exposed publicly, though it defaults to `127.0.0.1`. No `eval` or obfuscation is present.
Updated: 2026-01-16GitHub→
0
0
Low Cost
Jing-yilin icon
Sec8

Provides Model Context Protocol (MCP) access to Twitter data and write actions through the TwitterAPI.io service for LLM clients like Claude.

Setup Requirements

  • ⚠️Requires a `TWITTERAPI_API_KEY` from TwitterAPI.io (a paid service with free trial credits).
  • ⚠️Node.js version 18.0.0 or higher is required.
  • ⚠️Write actions (`login_user`, `create_tweet`) explicitly require a high-quality residential proxy.
Verified SafeView Analysis
The server securely handles API keys via environment variables and stores sensitive login cookies only in-memory. It uses HTTPS for all API requests and supports enterprise proxies. A moderate risk exists with client-provided `save_dir` arguments for file writes, as a malicious client could potentially attempt to write to arbitrary file paths on the local system if not properly restricted by the client-side sandbox; however, the content written is controlled (cleaned JSON data). No 'eval' or obvious obfuscation is present.
Updated: 2026-01-10GitHub→
0
0
Medium Cost
JustAnotherDevv icon

Arbitrum-Stylus-MCP

by JustAnotherDevv

Sec8

A privacy-preserving quadratic voting system leveraging Noir's zero-knowledge proofs (ZKPs) and Arbitrum Stylus smart contracts to verify voter eligibility, prevent double-voting, and calculate quadratic voting power with complete ballot secrecy.

Setup Requirements

  • ⚠️Requires `nsv` (Noir Stylus Verifier CLI) tool to be installed.
  • ⚠️Requires a specific Rust toolchain (e.g., `1.83.0` with `wasm32-unknown-unknown` target) for Stylus contract compilation.
  • ⚠️Requires `pnpm` for managing JavaScript dependencies.
Verified SafeView Analysis
The project adheres to standard practices for ZKP-based decentralized applications using Noir and Arbitrum Stylus. There is no presence of 'eval', obfuscation, or hardcoded sensitive secrets within the provided source code for deployment. Key management for `--private-key` during contract deployment is an operational security concern, not a code vulnerability. The core security relies on the correctness of the Noir circuit logic and the cryptographic primitives provided by the underlying Noir/Stylus frameworks.
Updated: 2025-11-28GitHub→
0
0
Low Cost

fastmcp-server

by swissmarley

Sec3

Local web GUI and MCP server for developing and experimenting with FastMCP-based tools, offering a web interface to manage functions and integrate with LLM clients.

Setup Requirements

  • ⚠️Requires the `fastmcp` package, which may not be a standard PyPI package and might require a specific installation method or source.
  • ⚠️Executing custom functions involves arbitrary Python code execution, which is a security risk; only trusted code should be used locally.
Review RequiredView Analysis
The server allows arbitrary Python code execution via the `POST /api/functions` endpoint, enabling users to add custom functions at runtime. While the project states it 'attempts to restrict builtins,' this is not a full sandbox and is explicitly warned as 'dangerous.' Running untrusted code is not recommended, and exposing the server beyond `localhost` would greatly increase the risk of remote code execution.
Updated: 2025-11-17GitHub→
0
0
Medium Cost
anortham icon

sherpa

by anortham

Sec9

Guides AI agents through systematic software development workflows, promoting best practices and habit formation via positive reinforcement.

Setup Requirements

  • ⚠️Requires Bun runtime (v1.0.0+) to be installed and in PATH.
  • ⚠️Requires a one-time setup step (`bun run setup`) to initialize user configuration and copy default workflows to `~/.sherpa/`.
  • ⚠️When configuring in Claude, the `sherpa-server.ts` path must be an absolute file path.
Verified SafeView Analysis
The server primarily operates locally, interacting via standard I/O (stdio) and persisting state to the user's home directory (`~/.sherpa/`). Input arguments are validated for length and type, and workflow YAML files are parsed, not executed. File operations for state persistence (`user-profile.json`, `progress-tracker.json`, `workflow-state.json`) use atomic writes (temp file + rename) and include robust error handling with retry mechanisms and graceful degradation. There are no obvious hardcoded secrets or direct network listening services exposed. The risk of `eval` or arbitrary code execution from workflow files is mitigated by parsing YAML rather than executing code. Overall, it demonstrates good security practices for a local agent-guidance tool.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
brownrl icon

eco_mcp

by brownrl

Sec8

Provides AI agents with access to the European Commission's Component Library (ECL) documentation, offering search, structured data, and code examples for frontend development.

Setup Requirements

  • ⚠️Requires a local SQLite database (`ecl-database.sqlite`) to be present and populated. This database is built by running the `crawl.js` script.
  • ⚠️Requires execution of the `download-ecl-assets.sh` script to fetch frontend assets (CSS, JS, icons) for local development if building pages from templates.
  • ⚠️Requires Node.js (version >=18 recommended based on `@modelcontextprotocol/sdk` dependencies).
Verified SafeView Analysis
The server-side Node.js code (`index.js`, `crawl.js`) uses parameterized SQL queries, mitigating SQL injection risks. It fetches data from known, official European Commission domains via HTTPS. No 'eval' or direct execution of arbitrary user-provided code is observed. The 'start_here' tool returns a shell script (`download-ecl-assets.sh`) as text content, which the user is then instructed to run manually; the server itself does not execute this script, placing the responsibility of reviewing and running the script on the user. The `ecl-ec.js` file is a minified client-side library and its direct impact on server security is minimal for this MCP.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
GitCheckoutMaster icon

Full_UI_Automation

by GitCheckoutMaster

Sec1

An AI assistant for Windows UI automation, specializing in file management and VS Code interactions, accessible via a chat interface.

Setup Requirements

  • ⚠️Requires a Windows operating system due to explicit calls to `powershell.exe` and hardcoded Windows paths for scripts.
  • ⚠️Requires Node.js runtime to be installed, as the `McpToolset` instances launch Node.js-based MCP servers.
  • ⚠️Requires a `GOOGLE_API_KEY` for the Gemini LLM, which must be provided via an `.env` file or environment variables.
Review RequiredView Analysis
CRITICAL: The `file_mgmt_server` and `vscode_mcp_server` tools utilize `node:child_process.exec` to run PowerShell scripts. User-provided `file_path` and `content` are directly interpolated into these commands (e.g., `-Path "${file_path}"`), creating severe command injection vulnerabilities. A malicious user could execute arbitrary code on the host system with the privileges of the running server. The use of `powershell.exe -ExecutionPolicy Bypass` further exacerbates this risk by disabling PowerShell's script execution restrictions. Additionally, the FastAPI endpoint `/api/chat` lacks any form of authentication or authorization, allowing any client to trigger these highly insecure operations. Hardcoded Windows paths also imply a lack of portability and potential misconfigurations if run in different environments.
Updated: 2025-11-27GitHub→
0
0
Low Cost
varun-dev-tech icon

mcp_server

by varun-dev-tech

Sec1

This repository appears to be a placeholder or initial setup for a server application, specifically named 'mcp server', but no functional code is provided.

Review RequiredView Analysis
CRITICAL: No source code beyond the README.md file was provided for analysis. Therefore, a comprehensive security audit is impossible. The score is extremely low due to the complete lack of visibility into potential vulnerabilities, malicious patterns, or hardcoded secrets. It is impossible to determine if 'eval' or other dangerous functions are used.
Updated: 2025-12-05GitHub→
0
0
Low Cost
nik1097 icon
Sec9

Estimates the token cost of individual tool responses from an MCP server to help clients understand LLM inference expenses.

Setup Requirements

  • ⚠️Requires Python 3.13 or newer.
  • ⚠️An existing MCP server is required for the tool to analyze.
  • ⚠️A Bearer token (`--token`) may be required by the target MCP server for authentication.
Verified SafeView Analysis
The project functions as a client to an external MCP server, with the server URL and authentication token provided by the user. It uses standard JSON parsing (`json.loads`) which is generally safe. No 'eval' or similar dangerous functions were found. The primary security consideration is operational: ensuring the user points the client to a trusted MCP server, as the tool will interact with whatever URL it is given.
Updated: 2025-12-11GitHub→
0
0
Medium Cost
MinatoNami icon

project-alena

by MinatoNami

Sec7

Orchestrates various AI models and tools (like Google Calendar and Codex CLI) via MCP for local, privacy-first personal assistant capabilities, integrating voice, chat, and external services like Telegram.

Setup Requirements

  • ⚠️Codex CLI Required: Requires the Codex CLI to be installed and configured, which may involve a paid subscription.
  • ⚠️Google OAuth Credentials: Manual setup in Google Cloud Console is required to obtain OAuth 2.0 credentials (`credentials.json`) for Google Calendar integration.
  • ⚠️Hardcoded `ALLOWED_REPOS`: The `modules/core/controller/safety.py` file contains a hardcoded whitelist of allowed repository paths (`/Users/lionelchong/sandbox`) for Codex operations, which must be manually modified for different development environments.
  • ⚠️Ollama Local Server: Requires a local Ollama instance running (default `http://localhost:11434`) with a language model downloaded (e.g., `llama3.1` or `gpt-oss:20b`).
Verified SafeView Analysis
Controlled Subprocess Execution: The `codex_runner.py` uses `subprocess.run` to invoke the `codex` CLI. While it attempts to sandbox with `--sandbox workspace-write` for `--apply` actions, any external CLI invocation carries inherent risks if the CLI itself or the prompt input can be exploited for arbitrary command execution. The project states "No shell execution (safe subprocess calls)" but this relies on the `codex` CLI's own sandboxing. Hardcoded Repository Whitelist: The `modules/core/controller/safety.py` module includes `ALLOWED_REPOS = ["/Users/lionelchong/sandbox"]` which is a hardcoded path. This is a critical security control to prevent agents from accessing arbitrary filesystem locations, but it must be made configurable by the user or dynamically determined to be practical and secure in different deployment environments. As it stands, it severely limits usability and is a significant setup friction. Default Wildcard CORS: The FastAPI backend for the voice assistant (`modules/voice-assistant/backend/app/main.py`) uses `allow_origins=["*"]` for CORS by default. While common for local development, this poses a security risk if the server is exposed to the internet, as it allows any domain to make cross-origin requests. Secrets Management: `modules/mcp/google-calendar/secrets/README.md` explicitly instructs users not to commit `credentials.json` or `token.json` and details secure handling, which is good practice. Environment variables are supported for custom paths.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
aryanguptacsvtu icon

test-remote-mcp-server

by aryanguptacsvtu

Sec8

Manages personal expenses by allowing users to add, list, and summarize financial transactions.

Setup Requirements

  • ⚠️Requires Python 3.12 or higher.
  • ⚠️Expense data is stored in a temporary SQLite database, meaning data will not persist across server restarts or system reboots.
  • ⚠️Database file permissions might cause issues if the temporary directory is not writable by the server process.
Verified SafeView Analysis
The server uses parameterized SQL queries, effectively mitigating SQL injection risks. No 'eval' or obvious hardcoded secrets are present. The database is stored in a temporary directory, which while not ideal for persistence, doesn't inherently introduce a security vulnerability beyond potential unauthorized access by other local processes in a shared environment (a general OS concern). The 'categories.json' is read from a local file, posing minimal risk unless the file system itself is compromised.
Updated: 2025-11-22GitHub→
0
0
Medium Cost
donjosh-ua icon

localmcp

by donjosh-ua

Sec5

Provides basic mathematical addition and CSV data analysis functionalities as an MCP server tool.

Setup Requirements

  • ⚠️Docker required
  • ⚠️MCP gateway required
  • ⚠️Python dependencies (pandas)
Review RequiredView Analysis
The `get_csv_stats` tool allows a `file_path` argument that directly uses `pd.read_csv()`. Without proper input validation or strict sandboxing (e.g., within a Docker volume mounted only to '/data' and strict path enforcement by the gateway), this could lead to arbitrary file disclosure within the container's accessible filesystem if a malicious path is provided.
Updated: 2025-11-29GitHub→
PreviousPage 472 of 760Next