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
marco110 icon
Sec3

Provides real-time vehicle data to an AI agent (Xiaozhi) via the Model Context Protocol (MCP) using a Node.js server.

Setup Requirements

  • ⚠️Hardcoded API login credentials (`loginName`, `loginPwd`) must be replaced with secure environment variables or a secrets management solution.
  • ⚠️A `token.js` file is required in the project root to provide the MCP endpoint token, and its content is not provided in the source.
  • ⚠️The external API endpoint `https://gateway-test.wuzheng.com.cn` is a 'test' gateway, which might have access restrictions or be unstable.
Review RequiredView Analysis
CRITICAL: The `getToken` function in `mcp-server.js` contains hardcoded login credentials (`loginName`, `loginPwd`) for an external API. The `AUTH_TOKEN` in `API_CONFIG` has a hardcoded default value ('your-authorization-token') if the `AUTH_TOKEN` environment variable is not set. The `index.js` file imports a token from `./token.js` which likely contains another hardcoded secret, though the file content isn't provided. These hardcoded secrets pose significant security risks.
Updated: 2025-12-06GitHub→
0
0
Medium Cost
abumalick icon

openapi-mcp

by abumalick

Sec8

Enables LLMs to explore OpenAPI specifications by loading, listing, and detailing API endpoints, parameters, and schemas.

Setup Requirements

  • ⚠️Requires Node.js 18+
  • ⚠️Only supports OpenAPI 3.x specifications (not Swagger 2.0)
  • ⚠️Potential for Server-Side Request Forgery (SSRF) if openapi_load is used with untrusted URLs in an insecure environment.
Verified SafeView Analysis
The server can fetch OpenAPI specifications from arbitrary URLs provided by the LLM, posing a potential Server-Side Request Forgery (SSRF) risk if not properly isolated or access-controlled in its deployment environment. Schemas are truncated to prevent excessive output, which also helps mitigate against very large (potentially malicious) schema definitions. No direct code execution vulnerabilities or hardcoded secrets were found.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
glasses666 icon
Sec8

A Python-based MCP server for image recognition leveraging various large language models to describe images or answer questions about them.

Setup Requirements

  • ⚠️Requires API Keys for commercial LLMs (Google Gemini, OpenAI, Alibaba Cloud DashScope, Volcengine Ark) which may incur costs.
  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Manual configuration of API keys in the `.env` file is necessary after setup.
  • ⚠️The server processes local file paths as image input; ensure proper access controls in a shared environment.
Verified SafeView Analysis
The server uses `os.getenv` for API keys, which is good practice. Image data can be sourced from URLs, Base64 strings, or local file paths. While convenient, allowing arbitrary local file paths as image input could pose a local file disclosure risk if the server is exposed to untrusted users without proper input validation or sandboxing. However, for its intended use as a local MCP server for agents, this risk is mitigated. No 'eval' or direct shell command execution found.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
htooayelwinict icon

open-bridge

by htooayelwinict

Sec9

A lightweight MCP server enabling AI coding assistants to interact with the OpenRouter API for various coding tasks.

Setup Requirements

  • ⚠️Requires an OpenRouter API Key, which typically involves usage-based costs from OpenRouter.
  • ⚠️Requires Python 3.10 or newer.
  • ⚠️The recommended installation method uses 'uvx', which might require an additional installation step for users unfamiliar with it.
Verified SafeView Analysis
The code avoids direct use of 'eval' or 'exec'. API keys are loaded from environment variables, which is a secure practice. It uses 'httpx' for async network calls, a standard and reliable library. The 'directory' parameter is only used for validation with 'os.path.isdir' and for metadata, not for command execution, mitigating path traversal risks. JSON extraction uses regex and 'json.loads' with error handling, preventing code injection through malformed JSON. The stateless nature further reduces potential attack surfaces.
Updated: 2026-01-18GitHub→
0
0
Low Cost
miriamtoca-ui icon

notion-mcp-server

by miriamtoca-ui

Sec4

This server acts as a Model Context Protocol (MCP) endpoint, enabling agents to interact with Notion for document management, including creating, retrieving, and updating entries, and provides a direct `/ingest` endpoint for data entry.

Setup Requirements

  • ⚠️Requires NOTION_TOKEN environment variable (from a Notion integration).
  • ⚠️Requires NOTION_DATABASE_ID environment variable (from a Notion database, critical to avoid using hardcoded default).
  • ⚠️Requires a Notion integration to be set up with appropriate permissions to the target database(s).
Review RequiredView Analysis
No 'eval' or obfuscation was found. Secrets (NOTION_TOKEN) are intended to be loaded from environment variables. However, a critical flaw exists in `tools.js`: `NOTION_DATABASE_ID` has a hardcoded default value ('2b704d5815c48030bf12f039d1a06893') which will be used if the environment variable is not explicitly set. Although `server-http.js` logs a warning, the server will proceed to operate on this potentially unintended or public database. This poses a significant risk of data exposure or accidental writes to the wrong database. The `/mcp/call_tool` endpoint dynamically executes handlers, relying on input schema validation and Notion's API to prevent misuse, but still represents a broad attack surface if tool arguments can be crafted maliciously. The `/ingest` endpoint also accepts arbitrary JSON for Notion page creation.
Updated: 2025-12-10GitHub→
0
0
Low Cost
like-a-freedom icon

rusty-intervals-mcp

by like-a-freedom

Sec9

An MCP server for Intervals.icu, enabling LLMs (like Claude and VSCode Copilot) to access and analyze athlete training data, wellness metrics, and performance.

Setup Requirements

  • ⚠️Requires a valid Intervals.icu API Key.
  • ⚠️Requires your Intervals.icu Athlete ID.
  • ⚠️Requires Rust 1.92+ and Cargo, or Docker, for execution.
Verified SafeView Analysis
The project demonstrates strong security practices: no hardcoded secrets (API keys and athlete IDs are environment variables), explicit input validation (date formats, empty queries), and robust webhook handling with HMAC-SHA256 verification and deduplication. The `CONTRIBUTING.md` reinforces security best practices by advising against storing secrets in the repository. Network security for remote deployment emphasizes TLS and an authentication layer, which is commendable.
Updated: 2026-01-18GitHub→
0
0
Low Cost
Angelxd0714 icon

mcp-payment-server

by Angelxd0714

Sec1

Processes and manages financial payments within a server environment.

Review RequiredView Analysis
Cannot perform a comprehensive security audit as the source code was truncated and not provided for analysis. Given the name 'mcp-payment-server', it likely handles highly sensitive financial data, making a thorough security review critical but impossible with the provided information. No specific risks like 'eval', hardcoded secrets, or malicious patterns can be identified without access to the code.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
lightningfaucet icon

mcp-server

by lightningfaucet

Sec8

Enables AI agents to manage Bitcoin wallets on the Lightning Network, supporting payments, invoicing, and L402 protocol access for AI economic autonomy.

Setup Requirements

  • ⚠️Requires a Lightning Faucet API key, which can be obtained via self-registration using the `register_operator` tool, or pre-configured as an environment variable.
  • ⚠️Relies on the external Lightning Faucet service (lightningfaucet.com) for all transaction processing and wallet management.
  • ⚠️Requires Node.js version 18.0.0 or higher to run.
Verified SafeView Analysis
The server uses Zod for input validation, mitigating common injection risks. API keys are sourced from environment variables, which is a good practice for secret management. There is no direct use of 'eval' or similar dynamic code execution functions. The client makes secure HTTPS calls to the external 'lightningfaucet.com' API. Past security fixes (e.g., error exposure, XSS in webhook URLs) indicate a history of addressing vulnerabilities, improving confidence in current versions. Key security is largely dependent on the user's handling of the API key and recovery codes.
Updated: 2026-01-18GitHub→
0
0
Low Cost
Sav1ouR520 icon
Sec8

Educational management for teachers, integrating with AI assistants via MCP to manage resources, create questions, and track tasks.

Setup Requirements

  • ⚠️Requires Python 3.11+
  • ⚠️Requires authentication credentials (XIAOYA_ACCOUNT/XIAOYA_PASSWORD or XIAOYA_AUTH_TOKEN) to be set via environment variables or HTTP headers.
  • ⚠️Specifically designed for the '小雅智能教学平台' (xiaoya intelligent teaching platform) due to hardcoded API endpoints and client IDs.
Verified SafeView Analysis
No use of 'eval' or obvious code obfuscation. Network communication uses HTTPS. Authentication involves fetching and caching tokens, which is standard. Hardcoded URLs (fzrjxy.ai-augmented.com) and client/school IDs tie the server to a specific platform, which isn't a vulnerability per se but limits general applicability. Input validation via Pydantic models is present for tool arguments. File download uses URL quoting, reducing injection risk. Overall appears robust for its intended purpose.
Updated: 2025-11-22GitHub→
0
0
Medium Cost
namphuongtran icon

mcp-server

by namphuongtran

Sec8

Provides a framework for managing, interacting with, and inspecting model context and related knowledge bases using the Model Context Protocol.

Setup Requirements

  • ⚠️Requires Docker and Docker Compose installed.
  • ⚠️Requires `OUTLINE_API_KEY` environment variable.
  • ⚠️Requires `MCP_PROXY_AUTH_TOKEN` environment variable.
Verified SafeView Analysis
No obvious security risks in the provided `docker-compose.yml`. Secrets (`OUTLINE_API_KEY`, `MCP_PROXY_AUTH_TOKEN`) are passed via environment variables, not hardcoded. The security of the application code within the Docker images is not assessable from this file.
Updated: 2025-11-30GitHub→
0
0
Low Cost
MagnusMax-tech icon

Kopi-af-MCP-server

by MagnusMax-tech

Sec7

A simple web application and API for students to view and sign up for extracurricular activities at a high school.

Setup Requirements

  • ⚠️Data is stored in-memory and will be reset when the server restarts.
  • ⚠️The provided `src/README.md` suggests running with `python app.py`, but a FastAPI application is typically run using Uvicorn, e.g., `uvicorn app:app --host 0.0.0.0 --port 8000` to start the server properly.
Verified SafeView Analysis
The application uses an in-memory database, which resets on server restart. No hardcoded secrets, 'eval', or obfuscation found. The frontend uses `innerHTML` to render activity details and participant emails. While not currently exploitable due to static data, dynamic user-controlled inputs rendered via `innerHTML` could pose an XSS risk. Lacks production-grade features like rate limiting, comprehensive input sanitization, and authentication/authorization.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
joelab1224 icon

mcp-server

by joelab1224

Sec1

A multi-tenant server designed to dynamically load, compile, and execute AI tools (like user profiling and text analysis) within a sandboxed environment, exposing them via an API.

Setup Requirements

  • ⚠️Requires MongoDB connection via MONGODB_URI and DATABASE_NAME environment variables.
  • ⚠️Default API_KEY and ADMIN_API_KEY environment variables are hardcoded ('dev-key-123', 'admin-key-123') and should be changed for production.
  • ⚠️Requires a modern Python version (likely 3.7+) due to asyncio usage.
Review RequiredView Analysis
CRITICAL VULNERABILITY: The `core/tool_compiler.py` explicitly allows `__import__` in the `safe_globals` dictionary passed to `exec`. This means a malicious tool can import and execute arbitrary Python modules (e.g., `os`, `sys`, `subprocess`) regardless of the `dangerous_patterns` regex, leading to remote code execution. This flaw is explicitly identified and warned against in `SECURITY_RECOMMENDATIONS.md` but not implemented in the provided compiler code. Additionally, hardcoded default values for `ADMIN_API_KEY` ('admin-key-123') and `API_KEY` ('dev-key-123') in environment variables pose a risk if not changed in production. The system lacks robust AST-based validation, bytecode inspection, resource monitoring, and subprocess isolation, all outlined as necessary improvements in the `SECURITY_RECOMMENDATIONS.md`.
Updated: 2025-11-27GitHub→
PreviousPage 385 of 760Next