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
Low Cost
ksrijansingh icon

mcp-server

by ksrijansingh

Sec7

Acts as an API gateway to expose MuleSoft services as 'tools' for consumption by external agents, potentially LLMs like Copilot, through a standardized interface.

Setup Requirements

  • ⚠️Requires a running MuleSoft API accessible at `MULE_BASE_URL` (defaults to `http://localhost:8081`)
  • ⚠️Node.js and npm are required to run the server.
Verified SafeView Analysis
The server acts as a proxy, directly forwarding client-provided payloads to configurable MuleSoft endpoints. While it validates `toolName`, it does not perform deep input validation of the `payload` against the defined JSON schemas. This means the security of the overall system heavily relies on the upstream MuleSoft APIs to sanitize and validate all inputs. An attacker could potentially send malicious payloads if the MuleSoft APIs are not robustly secured. `MULE_BASE_URL` is configurable via environment variable, which, if misconfigured to an untrusted or sensitive internal endpoint, could pose a risk. No `eval` or obvious obfuscation found.
Updated: 2025-12-03GitHub→
0
0
Medium Cost
Calvin-Francis icon

github-mcp-server-local

by Calvin-Francis

Sec9

Connects AI tools (agents, assistants, chatbots) directly to GitHub's platform to read repositories, manage issues and PRs, analyze code, and automate workflows through natural language interactions.

Setup Requirements

  • ⚠️Requires a GitHub Personal Access Token (PAT) with appropriate scopes for authentication.
  • ⚠️Requires Docker to be installed and running for the recommended local container-based setup, or a Go development environment for building from source.
  • ⚠️Needs an MCP-compatible host application (e.g., VS Code, Claude, Cursor) to expose its tools to an AI model.
Verified SafeView Analysis
The project extensively documents secure handling of GitHub Personal Access Tokens (PATs) via environment variables, explicitly advising against hardcoding. Input validation is performed on tool parameters, and response content (like issue/PR titles/bodies) is sanitized using `bluemonday` to mitigate injection risks. A 'lockdown mode' feature restricts content visibility in public repositories based on user push access, enhancing privacy and security. The system's reliance on GitHub's API permissions means access is inherently constrained by the authenticated user's privileges.
Updated: 2026-01-19GitHub→
0
0
High Cost
xebialabs-community icon

community-release-llm-integration

by xebialabs-community

Sec7

Facilitate AI-powered automation workflows in Digital.ai Release by integrating with LLMs and Model Context Protocol (MCP) servers.

Setup Requirements

  • ⚠️Requires Python 3 and Docker to run the development environment.
  • ⚠️Requires modifying the local `hosts` file (`/etc/hosts` or `C:\Windows\System32\drivers\etc\hosts`) with specific entries, which needs sudo/administrator permissions.
  • ⚠️Requires API keys for various LLM providers (e.g., Gemini, OpenAI, Digital.ai LLM) and potentially MCP servers (e.g., GitHub, Agility), which are often paid services.
Verified SafeView Analysis
The system utilizes API keys/tokens for various LLM and MCP providers, which are handled as secrets within Digital.ai Release configurations. The `llm_agent.py` component uses LangChain agents that can invoke tools on configured MCP servers. While this is the intended functionality, it introduces a potential risk for unintended actions if an agent is poorly prompted or if connected MCP servers expose overly permissive or dangerous tools without sufficient safeguards. The development environment setup script (`spin-remote-runner.sh`) hardcodes `admin:admin` for initial token fetching, which is a common practice for local development but would be a critical vulnerability in a production setup. There are no direct `eval` or obvious command injection points from user inputs to the Python code itself, but the agent's ability to orchestrate external tool calls requires careful consideration of the security posture of all integrated MCP servers.
Updated: 2025-11-23GitHub→
0
0
Low Cost
a607ernie icon

mcp-server-http

by a607ernie

Sec2

This server acts as a Model Context Protocol (MCP) server, providing various mock tools (weather, employee info, user info, product search, news, calculator) via a Streamable HTTP API for potential use by AI models.

Setup Requirements

  • ⚠️Requires Python 3.13 or higher.
  • ⚠️The `PORT` environment variable should be set, otherwise it defaults to 8741.
  • ⚠️The `calculate` tool uses `eval`, which poses a severe security risk if exposed to untrusted input.
Review RequiredView Analysis
CRITICAL: The `calculate` tool in `src/app/tools/products.py` uses `eval(expression)` directly on user-provided input. This allows arbitrary code execution and is an extreme security vulnerability. The code explicitly acknowledges this risk with a comment, but it is still present. Additionally, `enable_dns_rebinding_protection` is explicitly set to `False` in `src/settings.py`, which could be a security concern depending on the deployment scenario.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
Theprofitplatform icon

coolify-mcp-server

by Theprofitplatform

Sec9

Provides an advanced automation and monitoring interface for managing Coolify infrastructure resources, deployments, and configurations.

Setup Requirements

  • ⚠️Requires an existing and operational Coolify instance with an API accessible via COOLIFY_BASE_URL and a valid COOLIFY_TOKEN.
  • ⚠️For full functionality of tools like 'get_application_deployment_history' and 'get_deployment_logs', the server must be co-located with or have 'docker exec' access to the 'coolify-db' container.
  • ⚠️Qdrant vector database (QDRANT_HOST, QDRANT_API_KEY) is recommended for semantic search capabilities, with an optional OPENAI_API_KEY for superior embeddings.
  • ⚠️Docker and Node.js (v18+) runtime are prerequisites for running the server itself.
Verified SafeView Analysis
The project has a strong focus on security, having remediated critical vulnerabilities related to hardcoded credentials and tokens in version 1.0.2. It employs environment variables for sensitive data and includes robust command injection prevention in server command execution tools. API requests are rate-limited with retry logic, enhancing resilience.
Updated: 2025-12-10GitHub→
0
0
Medium Cost
Bhoopesh123 icon

python_mcp_servers

by Bhoopesh123

Sec2

Automate monitoring tasks by integrating with Prometheus for metric data retrieval, generating PromQL queries, and dynamically creating or updating Grafana dashboards, alongside triggering n8n workflows.

Setup Requirements

  • ⚠️Requires a local Prometheus instance running (default: http://localhost:9090).
  • ⚠️Requires a local Grafana instance running (default: http://localhost:3000).
  • ⚠️Requires an n8n instance running (default: http://localhost:5678).
  • ⚠️A Grafana API Key must be generated and, ideally, configured as an environment variable (currently hardcoded in the provided examples).
Review RequiredView Analysis
The Grafana API key is hardcoded directly in multiple Python source files (`main.py`, `main_grafana_prom_automated.py`, etc.), which is a critical security vulnerability. Additionally, the `trigger_n8n_get` tool accepts an arbitrary `webhook_url` as input, which exposes the server to Server-Side Request Forgery (SSRF) attacks or allows it to be used for port scanning internal networks.
Updated: 2025-12-04GitHub→
0
0
Low Cost
cocolizh icon

mcp-demo-2048

by cocolizh

Sec9

This server provides a web-based implementation of the classic 2048 game, accessible through a browser and using WebSockets for real-time game state updates.

Setup Requirements

  • ⚠️Requires Node.js 18+ and npm installed locally (if not using Docker)
  • ⚠️Docker is recommended for simplified deployment and execution.
Verified SafeView Analysis
The source code analysis reveals no use of 'eval', obfuscation, hardcoded secrets, or overtly malicious patterns. It utilizes standard Node.js libraries (Express, ws) for a simple web game. Network exposure on port 8080 is standard for a web application.
Updated: 2025-11-20GitHub→
0
0
High Cost
adityasingh0z3 icon

obsidian-mcp-server

by adityasingh0z3

Sec8

Enhances Obsidian note-taking with advanced vault management, graph analytics, and semantic search capabilities, acting as a tool for large language models.

Setup Requirements

  • ⚠️Requires `OBSIDIAN_API_KEY` environment variable, obtained from the Obsidian Local REST API plugin settings.
  • ⚠️Requires the Obsidian Local REST API plugin to be installed, enabled, and configured in Obsidian.
  • ⚠️Requires the Obsidian Smart Connections plugin and Research MCP Bridge plugin to be installed for semantic search tools to function.
  • ⚠️Graph analysis tools and pattern search (when implemented) require the `OBSIDIAN_VAULT_PATH` environment variable to be set.
  • ⚠️Requires Node.js version 18 or later.
Verified SafeView Analysis
The server loads sensitive API keys from environment variables (OBSIDIAN_API_KEY), which is good practice. It connects to the local Obsidian Local REST API. While default SSL verification (`verifySsl: false`) is disabled by default for the Obsidian REST API client and explicitly for the Smart Connections service (`rejectUnauthorized: false`), this typically pertains to self-signed certificates in a local development environment and doesn't expose a critical external network risk as the communication is generally localhost-bound. The server itself uses standard I/O (stdio) for communication, minimizing external network attack surface. There is no evidence of `eval` usage, obfuscation, or malicious patterns. The `patch_content` tool is commendably disabled due to known upstream bugs, preventing potential data corruption. Local file system access is required for graph tools via OBSIDIAN_VAULT_PATH, which is an inherent part of its functionality.
Updated: 2026-01-19GitHub→
0
0
Low Cost
joe-watkins icon

mcp-hatchery

by joe-watkins

Sec9

Scaffolds Model Context Protocol (MCP) servers in JavaScript or Python, offering local and cloud deployment options for AI assistant integration.

Setup Requirements

  • ⚠️Requires Node.js >= 18.0.0 for the CLI tool and JavaScript projects.
  • ⚠️Requires Python >= 3.8 for FastMCP (Python) projects.
  • ⚠️Remote deployment options (Netlify, Vercel, FastMCP Cloud) require corresponding accounts and GitHub integration.
Verified SafeView Analysis
The CLI tool itself primarily performs file system operations (creating directories and writing template files) based on user prompts. There are no indications of 'eval' usage, obfuscation, or hardcoded secrets within the provided source code for the scaffolding logic. The generated MCP servers (both JavaScript and Python) call tool handlers directly without dynamic code execution from external input, relying on predefined tool logic. The security of the *generated* server's custom tools will depend on the developer's implementation, but the scaffolding framework provides a secure base. Network risks are inherent to any server exposing an API, but the templates use standard web frameworks and include CORS headers where appropriate.
Updated: 2025-12-28GitHub→
0
0
High Cost
iceinvein icon
Sec7

Provides semantic search and code navigation capabilities to LLM agents by locally indexing codebases.

Setup Requirements

  • ⚠️Requires an initial download of approximately 300MB for the AI embedding model.
  • ⚠️Creates and uses significant local disk space (in a '.cimcp/' directory) for its SQLite, Tantivy, and LanceDB indexes and embedding cache.
Verified SafeView Analysis
The server binary is distributed via GitHub releases and installed by an npm script, relying on the integrity of the release process. The `fastembed` embedding backend downloads AI models from Hugging Face, introducing an external dependency. A local web UI feature can be enabled, exposing an HTTP server on `127.0.0.1:8787` by default. No direct 'eval' or obvious hardcoded secrets were found.
Updated: 2026-01-19GitHub→
0
0
Low Cost
yeison-liscano icon

http_mcp

by yeison-liscano

Sec9

Implements a lightweight server for the Model Context Protocol (MCP) over HTTP and STDIO, allowing exposure of Python functions as discoverable and executable remote tools and prompts.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Intended to be integrated with a Starlette or FastAPI application; requires manual setup of `Starlette.lifespan` for state management and `AuthenticationMiddleware` for authorization scopes.
  • ⚠️Specific dependency versions are locked (e.g., Pydantic==2.12.5, uvicorn==0.38.0, starlette==0.50.0) which may lead to dependency conflicts in larger projects.
Verified SafeView Analysis
The server uses Pydantic for robust data validation and serialization of JSON-RPC messages, mitigating common injection and malformed request vulnerabilities. It supports scope-based authorization via Starlette's authentication system, providing a strong access control mechanism. No direct use of dangerous functions like 'eval' or 'exec' was found. HTTP message size limits are enforced. The STDIO transport constructs a dummy request object with controlled headers, preventing direct user-supplied header injection. Overall, the design prioritizes secure handling of inputs and outputs within the MCP specification.
Updated: 2025-12-24GitHub→
0
0
Medium Cost
EricJujianZou icon

Anti-Soy

by EricJujianZou

Sec7

Automated GitHub profile analysis for assessing developer skills and detecting AI-generated code in hiring processes.

Setup Requirements

  • ⚠️Requires GITHUB_TOKEN environment variable for GitHub API access (GraphQL API calls).
  • ⚠️Requires GEMINI_API_KEY environment variable for Google Gemini LLM access, which is a paid service.
  • ⚠️Requires 'uv' for Python dependency management and virtual environment setup.
Verified SafeView Analysis
The server clones external GitHub repositories into a temporary directory and processes their content for analysis. While safeguards like file size limits (`MAX_FILE_SIZE`, `MAX_TOTAL_CONTENT`) are in place, processing arbitrary external code always carries inherent risks (e.g., resource exhaustion from specially crafted large or malicious files). Subprocess calls for 'git clone' and 'git log' are constructed with parsed URL components and safe paths, mitigating direct command injection risks. CORS is set to 'allow_origins=["*"]' in 'server/main.py', which is insecure for production but specified in the README as a development setting. No direct 'eval' or obvious code execution from uncontrolled user input was found.
Updated: 2026-01-18GitHub→
PreviousPage 333 of 713Next