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
ankitpokhrel08 icon

MCP-Servers

by ankitpokhrel08

Sec6

A multi-tool chatbot connecting specialized servers for math, expense management, and animation creation, driven by an LLM.

Setup Requirements

  • ⚠️Requires a Google Gemini API Key (GOOGLE_API_KEY or GEMINI_API_KEY) which may incur costs depending on usage.
  • ⚠️Requires manual adjustment of hardcoded absolute file paths for `uv`, `python3`, `math_server.py`, and `manim_server.py` in the `SERVERS` configuration within `client/gemini_client.py` and `streamlit_app.py` to match the local environment.
  • ⚠️Requires cloning a separate `manim-mcp-server` repository and installing Manim for animation functionality, followed by setting the `MANIM_EXECUTABLE` path.
Review RequiredView Analysis
The application relies on hardcoded absolute file paths for local server executables (uv, python3) and server scripts (math_server.py, manim_server.py) which are specific to the developer's system, causing significant setup friction and potential risks if these paths point to malicious or non-existent executables on another system. It depends on a separately cloned Manim server (manim-mcp-server) whose code is not provided for audit. The remote 'expense' server introduces an external dependency with its own security profile. While the provided `math_server.py` tools appear to handle inputs safely with type hints, general execution of external processes requires careful consideration.
Updated: 2025-11-28GitHub→
0
0
Medium Cost
Gustavohsv icon

server-mcp

by Gustavohsv

Sec5

Provides a server implementation for the Minecraft Protocol, allowing clients to connect and interact within a game world for multiplayer gameplay.

Setup Requirements

  • ⚠️Requires Java Runtime Environment (JRE) to be installed.
  • ⚠️Needs specific port forwarding (e.g., 25565) configured on the host network or router to be accessible externally.
Review RequiredView Analysis
No source code content was provided in the prompt for analysis, making a thorough security audit impossible. General risks for a network server include unhandled input validation, authentication bypasses, denial-of-service vulnerabilities, and potential information disclosure. Specific vulnerabilities like 'eval' or hardcoded secrets could not be checked.
Updated: 2026-01-19GitHub→
0
0
Low Cost
sanketshivale icon

MCP-SERVER

by sanketshivale

Sec6

Provides an API to query a local file-based JSON database via the Model Context Protocol, exposing data retrieval and filtering tools.

Setup Requirements

  • ⚠️Requires a Node.js environment to run.
  • ⚠️Requires a 'db.json' file to exist in the server's working directory for data storage.
  • ⚠️The 'mcp.json' configuration file contains a hardcoded absolute path to the 'mcp-server.js' script which needs to be updated by the user for their specific environment.
Review RequiredView Analysis
The server directly accesses properties of the loaded JSON database object using user-provided 'table' and 'field' values (e.g., db[table], r[field]). While inputSchema is defined for these tools, the handler functions themselves do not perform explicit sanitization or restrictive type checking on these inputs. If an attacker can provide values like '__proto__' or 'constructor' for 'table' or 'field' (e.g., by bypassing or exploiting a flaw in the Model Context Protocol SDK's input validation), it could potentially lead to prototype pollution, arbitrary object property manipulation, or unexpected behavior. This poses a significant security risk if the server is exposed to untrusted input without robust external validation. Additionally, the server reads the entire 'db.json' file on every tool call, which could be a performance bottleneck for large files.
Updated: 2025-11-19GitHub→
0
0
Low Cost
neuhausjulian icon

vibe-coding_mcp-server

by neuhausjulian

Sec9

Minimal MCP server prototype for validating architecture, testing strategy, authorization concepts, and dev-tool support through a single user profile retrieval tool.

Setup Requirements

  • ⚠️Requires Node.js 18+ with ES modules support.
  • ⚠️Server startup requires specific environment variables for authorization (either DEV_STATIC_AUTHORIZATION and DEV_STATIC_AUTHORIZATION_USER_ID for dev, or JWT_JWKS_URL and JWT_EXPECTED_AUDIENCE for production). Failure to configure these will prevent the server from starting.
  • ⚠️Production mode requires an external JWT authentication provider (JWKS endpoint and audience).
Verified SafeView Analysis
The server employs robust authorization configuration validation at startup, clearly separating development (static ID) and production (JWT) modes. JWT validation uses standard libraries (jsonwebtoken, jwks-rsa) and fetches JWKS from a configured URL, which is a common and secure practice. No 'eval' or other directly dangerous patterns are present. Sensitive configurations are handled via environment variables. The system correctly expects JWT tokens as tool arguments in production, not environment variables.
Updated: 2025-11-22GitHub→
0
0
Low Cost
AGDevX icon
Sec9

Enables AI assistants to interact with arbitrary OpenAPI-defined APIs by dynamically generating Model Context Protocol (MCP) tools.

Setup Requirements

  • ⚠️Requires Node.js 18.0.0 or higher.
  • ⚠️Requires an OpenAPI 3.0/3.1 compatible API specification URL(s) for the target API.
  • ⚠️For development environments using self-signed TLS certificates on the target API, the `NODE_TLS_REJECT_UNAUTHORIZED=0` environment variable may be required.
Verified SafeView Analysis
The server transparently allows disabling TLS certificate verification (`NODE_TLS_REJECT_UNAUTHORIZED=0`) for environments with self-signed certificates. While this introduces a configurable security trade-off, it is explicitly documented and logged with warnings. File system operations for managing MCP client configurations are restricted to user-specific directories and include backup functionality. The server incorporates rate limiting to protect external APIs. No direct use of `eval`, obfuscation, or other critical malicious patterns was detected in the provided source code.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
Arpit-saxena-2004 icon

github_mcp_server

by Arpit-saxena-2004

Sec8

A lightweight Model Context Protocol (MCP) server that enables Claude Desktop to perform various GitHub actions like repository, branch, pull request, issue, and file management.

Setup Requirements

  • ⚠️Requires a GitHub Classic Personal Access Token (PAT) with `repo` and `delete_repo` scopes.
  • ⚠️Python 3.10+ is required.
Verified SafeView Analysis
The server relies on a highly privileged GitHub Classic Personal Access Token (PAT) with `repo` and `delete_repo` scopes. Users must understand that this grants the server extensive control over their GitHub repositories, including the ability to delete them. While this is explicitly documented and required for the server's functionality, it necessitates careful management of the PAT by the user. There are no obvious code-level vulnerabilities like 'eval' or unvalidated shell commands.
Updated: 2025-12-11GitHub→
0
0
Low Cost
mukhomedzyanov icon

ticktick-http-mcp

by mukhomedzyanov

Sec8

Provides a simple HTTP API wrapper for specific TickTick functionalities, acting as a Micro-Capability Platform (MCP).

Setup Requirements

  • ⚠️Requires a TickTick API Access Token (TICKTICK_ACCESS_TOKEN) to interact with the TickTick API.
  • ⚠️Requires Node.js 18.x as specified in package.json.
Verified SafeView Analysis
The server uses `Access-Control-Allow-Origin: '*'` which allows any origin to access the API. While suitable for some internal or development contexts, for a publicly exposed API, this is a security consideration. `TICKTICK_ACCESS_TOKEN` is loaded from environment variables, which is good practice. No 'eval' or obvious malicious patterns found.
Updated: 2025-11-23GitHub→
0
0
Low Cost
XuYunlei icon

multiagent-mcp

by XuYunlei

Sec8

A multi-agent customer service system that routes queries, retrieves customer data, manages support tickets, and coordinates multi-step workflows.

Setup Requirements

  • ⚠️Requires `langgraph` and `langchain-core` installation for full LangGraph SDK integration (optional fallback to simpler routing if not installed).
  • ⚠️Requires SQLite database initialization via `scripts/setup_database.py` before running services.
  • ⚠️For full independent agent functionality, requires starting multiple Python services concurrently (MCP Server, Customer Data Agent, Support Agent, Router Agent, Main Server), though a 'single process mode' is available for development.
Verified SafeView Analysis
The MCP HTTP Server's `call_tool` methods, particularly `update_customer`, use parameterized SQL queries and whitelisting for column names, effectively mitigating common SQL injection risks for both values and column identifiers. However, the system is designed for local/internal use, and it lacks authentication and authorization mechanisms for its HTTP endpoints and agent communication. The use of `allow_origins=["*"]` for CORS is also insecure for a publicly exposed production environment without proper access control.
Updated: 2025-12-02GitHub→
0
0
Low Cost
Traia-IO icon
Sec8

Provides AI agents and LLMs with tools to interact with the Nikola Test 2 API via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires Python 3.12 or newer.
  • ⚠️Requires Docker or Docker Compose for recommended setup.
  • ⚠️Requires D402 payment configuration (SERVER_ADDRESS, MCP_OPERATOR_PRIVATE_KEY, D402_FACILITATOR_URL) unless `D402_TESTING_MODE` is true. These can be generated automatically for local development by `run_local_docker.sh`.
Verified SafeView Analysis
The server uses the D402 payment protocol, expecting `SERVER_ADDRESS` and `MCP_OPERATOR_PRIVATE_KEY` via environment variables. These are generated for local testing by `run_local_docker.sh` if not already set. It exposes a public HTTP endpoint and uses CORS `allow_origins=["*"]`, which permits all origins. No `eval` or obvious malicious patterns were found in the provided source code. The server relies on the security of the `traia-iatp` dependency for D402 payment processing.
Updated: 2025-12-10GitHub→
0
0
Medium Cost
lksrpp icon
Sec8

Enables LLMs to interact with SQLite databases through read-only tools for schema inspection and SQL query execution.

Setup Requirements

  • ⚠️Requires Python 3.13+
  • ⚠️Requires 'uv' package manager for installation and execution
  • ⚠️The 'crm.db' database file must exist in the script's directory, or DB_PATH constant must be modified. Use 'uv run seed_database.py' to create it.
Verified SafeView Analysis
The server implements robust validation for the 'query' tool, explicitly checking that queries start with 'SELECT' or 'WITH' and blocking a comprehensive list of forbidden keywords (e.g., INSERT, UPDATE, DELETE, DROP) using word boundaries to prevent write operations and common SQL injection vectors. The 'describe_table' tool also pre-validates table names against 'sqlite_master' using parameterized queries before executing PRAGMA commands. While the 'query' tool executes raw SQL (after validation) rather than using parameterized queries for user-supplied SQL, the strict read-only enforcement significantly mitigates major risks.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
NonmaskableInt icon

alphavantage-mcp

by NonmaskableInt

Sec9

Provides comprehensive market data and fundamental analysis through the AlphaVantage API for integration with MCP-compatible clients like Claude Desktop.

Setup Requirements

  • ⚠️Requires a free AlphaVantage API key, which has a strict rate limit of 25 requests per day for the free tier.
  • ⚠️Requires the 'uv' package manager to be installed for dependency management and execution.
  • ⚠️Requires Python 3.10, 3.11, or 3.12.
Verified SafeView Analysis
The server follows good security practices by requiring the AlphaVantage API key via environment variables or a .env file, preventing hardcoding. Input validation functions are used for symbols, and Pydantic models enforce type and structure validation for other parameters. The `httpx` library is used for secure asynchronous HTTP requests. There are no indications of 'eval' or other dangerous patterns, and dependencies are standard and well-maintained.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
dreamwithrua icon

mcp-setting-all-server

by dreamwithrua

Sec8

This is a TypeScript Model Context Protocol (MCP) server boilerplate designed for rapid development of custom AI-powered services, including tools for basic utilities, image generation, and code analysis, which can be connected to clients like Cursor.

Setup Requirements

  • ⚠️Requires a Hugging Face API Token (hfToken) for image generation, otherwise the tool is not registered.
  • ⚠️Development and build workflows rely on the Smithery CLI (`npx smithery build`, `npx smithery dev`).
  • ⚠️Connecting to Cursor MCP requires an absolute path to the built server in `./.cursor/mcp.json`.
Verified SafeView Analysis
The server uses Zod for schema validation, which is a good practice. It handles a Hugging Face API token through configuration, preventing hardcoding. Operations are well-defined (calculator, greeting, time, server info, image generation, code review prompt). No direct arbitrary code execution or filesystem access from external requests beyond specified functionalities is observed. The 'code_review' prompt sends user-provided code to an LLM, which carries inherent prompt injection risks if not carefully managed by the client application consuming the LLM's output. Large image payloads from the 'generate_image' tool could consume memory, but this is a resource management concern rather than a direct vulnerability in the server logic.
Updated: 2025-11-27GitHub→
PreviousPage 694 of 713Next