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

MCP

by boham97

Sec4

This project provides a Multi-Client Protocol (MCP) server designed to enable Language Model Models (LLMs) to interact with external services and code execution environments (e.g., C, Java, Ruby, Python).

Setup Requirements

  • ⚠️Requires Ollama to be running locally with a compatible model (e.g., `qwen2.5:7b`).
  • ⚠️Requires manual configuration of MCP server connection details within `config.json`.
  • ⚠️Requires Python 3.10+ and pip version 25.1.1 or higher for client installation.
Review RequiredView Analysis
The system design allows Language Model Models (LLMs) to interact with external services and potentially trigger code execution (e.g., Ruby, Python scripts, C/Java environments). Without robust sandboxing, input sanitization, and privilege separation, this poses a significant risk of arbitrary code execution, command injection, or unauthorized resource access driven by malicious LLM outputs.
Updated: 2025-11-17GitHub→
0
0
High Cost
fwornle icon
Sec7

Performs comprehensive semantic analysis of code repositories and development activities, generating patterns and insights to build a unified knowledge base.

Setup Requirements

  • ⚠️Requires at least one LLM API key (e.g., GROQ_API_KEY, ANTHROPIC_API_KEY, OPENAI_API_KEY, GOOGLE_API_KEY) or a local Ollama instance running.
  • ⚠️Requires PlantUML executable for diagram generation.
  • ⚠️Requires Memgraph database and the `uv` (code-graph-rag) command-line tool for advanced code graph features.
Review RequiredView Analysis
The server makes extensive use of file system operations (`fs` module) and executes child processes (`git`, `plantuml`, `node workflow-runner.js`, `uv` for code-graph-rag). This is inherent to its function (analyzing local repositories) but implies significant privileges. If `repositoryPath` or workflow definitions are manipulated, this could lead to privilege escalation or arbitrary code execution. It relies on external LLM APIs and local services like Ollama/Memgraph, which may have their own security considerations (rate limiting, data privacy, unexpected responses). Internal workflow conditions use `new Function()` (similar to `eval`), which is generally risky but used within trusted configuration.
Updated: 2026-01-15GitHub→
0
0
Low Cost
Infogain-GenAI icon

repo-from-remote-mcp1

by Infogain-GenAI

Sec10

This repository appears to be a placeholder or testbed for an MCP server development, primarily containing testing documentation rather than executable server code.

Verified SafeView Analysis
The provided source code consists only of markdown files with 'testing' content. There are no executable components, 'eval' statements, obfuscation, network risks, hardcoded secrets, or malicious patterns detected, making it inherently safe.
Updated: 2025-12-02GitHub→
0
0
High Cost
GarviChittora icon

mcp-server-browserbase

by GarviChittora

Sec8

This server provides cloud browser automation capabilities, enabling LLMs to interact with web pages, extract data, take screenshots, and perform automated actions.

Setup Requirements

  • ⚠️Requires Browserbase API Key (paid service) and Project ID.
  • ⚠️Requires an LLM API key (e.g., GEMINI_API_KEY for default, or specific keys like ANTHROPIC_API_KEY, OPENAI_API_KEY for custom models), which are often paid services.
  • ⚠️Experimental features are available but may change or be removed in future releases.
  • ⚠️For SHTTP transport, binding to '0.0.0.0' requires careful network security considerations.
Verified SafeView Analysis
The server uses environment variables for API keys, which is a good practice. It explicitly checks for required API keys upon startup or session creation. However, the configuration allows binding to '0.0.0.0' for SHTTP transport, which should be used with caution as it exposes the server to all network interfaces. Initial config resolution in `src/config.ts` might assign dummy API keys if env vars are missing, though subsequent checks in `src/index.ts` and `src/sessionManager.ts` prevent actual functionality without valid keys. No direct `eval` or obvious malicious patterns were found in the provided code.
Updated: 2025-11-30GitHub→
0
0
Low Cost
sgasser icon

podcast-mcp

by sgasser

Sec9

Generates multi-speaker podcasts with AI voices directly inside Claude Desktop, processing all audio locally.

Setup Requirements

  • ⚠️Requires `uv` (a Python package manager) to be installed via Homebrew (on macOS/Linux) for Claude Desktop to find it.
  • ⚠️The first time it runs, it downloads the Coqui XTTS-v2 AI model (~2GB), which requires an internet connection and takes time.
  • ⚠️Requires Python 3.10-3.12.
Verified SafeView Analysis
The server operates entirely locally, minimizing network exposure. It uses `os.path.basename` for filename sanitization to prevent path traversal vulnerabilities. No use of `eval` or obfuscation was found. Temporary files are generally cleaned up in a `finally` block, though there's a minor edge case where a partially generated file might not be tracked for cleanup if the TTS generation fails before its path is added to the cleanup list.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
DecisionsDev icon
Sec3

The IBM ODM Management MCP Server acts as a bridge, exposing IBM ODM Decision Center REST API endpoints as tools for modern AI assistants and orchestration platforms.

Setup Requirements

  • ⚠️Requires Python 3.13 or higher, which may be newer than typically installed Python versions.
  • ⚠️Requires the `uv` Python package installer, which is less common than `pip`.
  • ⚠️Requires an operational IBM ODM Decision Center instance for full functionality, which may need to be set up separately (e.g., using Docker/Rancher Desktop or a remote instance).
Review RequiredView Analysis
The `CustomHTTPAdapter` explicitly disables hostname verification (`assert_hostname = False`) when making HTTPS requests, even if SSL verification is enabled (`verify_ssl=True`). This significantly weakens TLS security, making the server vulnerable to Man-in-the-Middle (MITM) attacks if an attacker can present a valid certificate from a trusted Certificate Authority (or a configured custom certificate) that does not match the target hostname. While enabling `--verifyssl "False"` is an explicit override for development, the default behavior with `assert_hostname = False` introduces a critical vulnerability for general use. Sensitive credentials are generally handled via environment variables, which is good practice.
Updated: 2026-01-19GitHub→
0
0
Low Cost
Lockpicher icon

mcp-n8n-server

by Lockpicher

Sec1

Based on the repository name mcp-n8n-server, this server likely provides an integration layer for n8n workflows to interact with a Multi-Chain Protocol (MCP), enabling automation of blockchain-related tasks.

Setup Requirements

  • ⚠️Requires an n8n instance for integration
Review RequiredView Analysis
No source code was provided, therefore a security audit cannot be performed. It is impossible to assess the presence of 'eval', obfuscation, hardcoded secrets, or other malicious patterns without the code.
Updated: 2025-12-01GitHub→
0
0
Medium Cost

learn-langchain-js-ai

by devravinder

Sec7

A learning and experimentation repository for understanding and implementing AI applications using Langchain.js and Langgraph.js.

Setup Requirements

  • ⚠️Requires API keys for Large Language Models (e.g., OpenAI, Anthropic, Google) configured in a `.env` file, as indicated by `.env.sample`.
  • ⚠️Node.js runtime and pnpm package manager are required to set up and run the project.
  • ⚠️Specific examples might require a MongoDB Atlas instance for state management or vector store, as suggested by referenced YouTube content.
Verified SafeView Analysis
Unable to inspect code for 'eval' or obfuscation due to truncated file access. Assumes standard security practices for a learning project. Relies on environment variables (via .env.sample) for API keys, which is standard. External network risks are primarily associated with calls to LLM APIs and other services as configured.
Updated: 2025-11-17GitHub→
0
0
High Cost
Novathegoat1 icon

lanhu-mcp

by Novathegoat1

Sec8

Integrate Lanhu design documents and team collaboration with AI IDEs for automated requirement analysis, UI design viewing, slice extraction, and context sharing.

Setup Requirements

  • ⚠️Requires manual acquisition of a Lanhu account cookie (`LANHU_COOKIE`) from browser developer tools.
  • ⚠️Requires Python 3.10+ and Playwright browser (Chromium) to be installed.
  • ⚠️The server binds to `0.0.0.0` by default; requires proper firewall rules or reverse proxy if exposed to a public network.
Verified SafeView Analysis
The server uses `LANHU_COOKIE` for authentication with Lanhu, which is a sensitive credential. While the project correctly advises storing this in environment variables and provides clear guides on securing it (`.env` file permissions, not committing to Git, regular rotation), user negligence in handling this cookie or exposing the server externally without proper network security (e.g., firewalls, HTTPS for production) could lead to risks. The project explicitly states that `0.0.0.0` is the default host and advises securing it. Input validation for `search_regex` is done via `re.compile`, which could be a ReDoS vector if not properly sanitized, though it operates on internal message content. File system operations are confined to the `./data` directory. No arbitrary code execution from user input or malicious patterns were found. The presence of `SECURITY.md` and detailed deployment best practices demonstrates a proactive security posture.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
Mitalee89 icon

n8n-mcp-workflow

by Mitalee89

Sec7

An AI-powered assistant for managing software development workflows in Jira and GitHub, querying Supabase data, and interacting via Slack.

Setup Requirements

  • ⚠️Requires an n8n instance to host and run these workflows.
  • ⚠️Requires configured API credentials for Jira Software Cloud, GitHub (OAuth2), Supabase, Slack, and Google Gemini (PaLM).
  • ⚠️The MCP Client node's endpoint is hardcoded to a specific n8n cloud URL (`https://workflowsforlife.app.n8n.cloud/mcp/7ad7d262-638c-456e-b88a-1aad3b562eb5`), which must be adjusted if deploying to a different n8n instance (e.g., local or self-hosted).
  • ⚠️Effective operation requires careful prompt engineering to prevent AI agent misuse via prompt injection.
Verified SafeView Analysis
The system acts as an AI agent using user input from Slack (via `{{ $json.text }}`) to control tools for Jira, GitHub, and Supabase. This pattern introduces a significant risk of 'prompt injection,' where malicious user input could manipulate the AI to execute unintended or harmful operations through the connected tools (e.g., creating spurious issues, altering data). While n8n handles credentials securely, the trust placed in AI-generated parameters for tool execution requires robust prompt engineering and input validation. The `MCP Client` also relies on a hardcoded n8n cloud endpoint for the `MCP Server`, creating a dependency on an external, potentially configurable, resource.
Updated: 2025-12-10GitHub→
0
0
Low Cost
ASMAMAWABATEH icon

mcp-dual-server-demo

by ASMAMAWABATEH

Sec9

Demonstrates a multi-transport (STDIO and HTTP) tool-calling server and client architecture using FastMCP for Python applications.

Setup Requirements

  • ⚠️Requires `fastmcp` package to be installed.
  • ⚠️Requires two separate terminal instances for server and client scripts to run the full demo.
Verified SafeView Analysis
The HTTP server binds to localhost (127.0.0.1), limiting direct external exposure. No dangerous functions like `eval` or hardcoded secrets found. The tool implementations are simple arithmetic and string manipulations and pose no apparent security risks.
Updated: 2025-12-02GitHub→
0
0
Low Cost
davidakpele icon

linkwarden-mcp-server

by davidakpele

Sec9

Provides AI assistants with programmatic access to Linkwarden instances for bookmark management, collections, links, and search functionality.

Setup Requirements

  • ⚠️Requires access to an existing Linkwarden instance (self-hosted or cloud).
  • ⚠️Requires a Linkwarden API token with appropriate permissions for the enabled toolsets.
  • ⚠️Requires Docker for the recommended installation method, or Go 1.23+ for building from source.
Verified SafeView Analysis
The server correctly uses environment variables or config files for API tokens, avoiding hardcoded secrets. It implements standard bearer token authentication for Linkwarden API calls. There are no 'eval' or obfuscation patterns observed. It supports an optional read-only mode for enhanced safety. Potential risks are tied to the security of the underlying Linkwarden instance and the scope of the provided API token.
Updated: 2025-12-01GitHub→
PreviousPage 422 of 760Next