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

python-mcp-server

by changren-wcr

Sec9

Deploys a simple tool-serving API using fastmcp that exposes basic arithmetic functions, primarily intended for integration with AI agents or other systems requiring callable external tools.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Project dependencies (fastmcp, uvicorn) must be installed.
  • ⚠️Intended for deployment (e.g., Google Cloud Run) but can be run locally.
Verified SafeView Analysis
The provided source code is clean, with no apparent malicious patterns, hardcoded secrets, or dangerous functions like 'eval'. It serves a simple API; overall security relies on the FastMCP library and the deployment environment's network configuration.
Updated: 2025-11-25GitHub→
0
0
High Cost
cpatel1293 icon

PRD-MCP-Server

by cpatel1293

Sec1

A specialized Model Context Protocol (MCP) server dedicated to creating and validating Product Requirements Documents (PRDs) using AI and customizable templates.

Setup Requirements

  • ⚠️Requires API keys for AI providers (e.g., OPENAI_API_KEY), which are typically paid services.
  • ⚠️Google Gemini, Anthropic Claude, and Local Model providers are currently stubbed implementations and will throw 'not yet implemented' errors if selected.
Review RequiredView Analysis
Critical vulnerability: The `get_provider_config` MCP tool exposes all configured AI provider API keys (e.g., OPENAI_API_KEY, ANTHROPIC_API_KEY) to any connected MCP client. This means anyone who can connect to the server can steal these secrets. High severity vulnerability: Multiple tools (`get_logs`, `export_templates`, `import_templates`) are vulnerable to path traversal attacks, allowing arbitrary file read/write on the server's filesystem by manipulating input parameters like `fileName` or `filePath`.
Updated: 2025-11-20GitHub→
0
0
Medium Cost
Dynava icon
Sec8

This server provides an MCP interface to interact with a Genesys transcription API, enabling the retrieval, generation, and querying of conversation transcripts.

Setup Requirements

  • ⚠️Requires the 'HOST' environment variable to be set to the transcription API endpoint, defaulting to 'http://9.223.220.100' if not provided.
  • ⚠️Requires Python 3.12 or newer.
  • ⚠️The backend Genesys API requires authentication for certain operations (e.g., 'interval_conversations').
Verified SafeView Analysis
The server acts as a proxy, forwarding requests to an external API defined by the HOST environment variable. While no direct 'eval' or hardcoded secrets are present, the security largely depends on the trustworthiness and security of the backend transcription API it connects to. Improper configuration of the HOST variable could lead to connections to untrusted services. The direct passing of arguments to the backend API could expose it to malicious input if the backend isn't robustly validated.
Updated: 2025-12-02GitHub→
0
0
Medium Cost
pradhyu icon
Sec8

Provides sample implementations for a Minimal Chat Protocol (MCP) server, either via FastAPI for HTTP or stdio for subprocess communication, using LangChain with OpenAI-compatible APIs.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid)
  • ⚠️Docker required
  • ⚠️Python 3.10+
Verified SafeView Analysis
The code itself avoids dangerous patterns like `eval` or hardcoded secrets. It relies on environment variables for API keys, which the READMEs correctly advise against for production in favor of secrets management. The FastAPI version exposes a standard HTTP port. The stdio version communicates via stdin/stdout, which is generally safer for isolated process communication. Production hardening (like multi-stage builds, healthchecks, robust framing) is recommended but not handled by the sample code.
Updated: 2025-11-25GitHub→
0
0
Medium Cost
a1loy icon

gerrit-mcp

by a1loy

Sec8

Provides an MCP (Model Context Protocol) interface to query a Gerrit instance for code review changes and project information, acting as a tool for potential AI agents.

Setup Requirements

  • ⚠️Requires access to a Gerrit instance, defaulting to `https://chromium-review.googlesource.com` unless specified via `-gerrit-instance`.
  • ⚠️Gerrit authentication, if enabled with `-with-auth`, necessitates specific environment variables such as `GERRIT_USERNAME` and `GERRIT_PASSWORD` (for basic/digest) or `GERRIT_COOKIE_NAME` and `GERRIT_COOKIE_VALUE` (for cookie auth).
  • ⚠️The MCP server's own authentication (via Bearer header) requires the `BEARER_TOKEN` environment variable to be set for the server to validate client requests.
Verified SafeView Analysis
The server employs environment variables (e.g., BEARER_TOKEN, GERRIT_USERNAME, GERRIT_PASSWORD) for handling sensitive authentication credentials, which is a good security practice. There is no evidence of 'eval' or other dynamic code execution patterns. Input parsing for tool requests leverages `mcp.ParseString` and `mcp.ParseInt`, and interactions with Gerrit are managed through a dedicated Go client library. URL validation includes hostname checks to prevent queries against unauthorized Gerrit instances. While `panic` calls exist for critical initialization failures, they are not within runtime request handling paths, thus not directly posing a runtime security flaw, but could impact server robustness. Overall, the code demonstrates reasonable security hygiene for its function.
Updated: 2025-11-30GitHub→
0
0
Low Cost
omnifyjp icon
Sec9

Enables AI assistants like Claude to create and validate Omnify schemas, facilitating structured data modeling.

Setup Requirements

  • ⚠️Requires Node.js and npm/npx to run.
  • ⚠️Requires configuration within Claude Code/Desktop's MCP settings for direct integration.
Verified SafeView Analysis
The server primarily processes structured text input (schema definitions) and generates text output (YAML, markdown). It uses standard libraries like 'yaml' and 'zod' for parsing and validation. No 'eval' or direct execution of untrusted code is observed. The scope is well-defined, mitigating common web vulnerabilities, and no hardcoded secrets or direct network interactions beyond stdio are present.
Updated: 2026-01-19GitHub→
0
0
Low Cost
gilamran icon
Sec9

A minimal MCP server template demonstrating both STDIO and HTTP transports with an example tool for filtering an in-memory TV catalog.

Setup Requirements

  • ⚠️To run the server in HTTP mode, you must create a '.env' file and set `MCP_SERVER_TRANSPORT_TYPE=HTTP`.
  • ⚠️For public access to the HTTP server, a reverse proxy (e.g., ngrok) is required.
  • ⚠️Requires Node.js and Yarn for initial setup (installing dependencies and building).
Verified SafeView Analysis
No 'eval' or code obfuscation found. No hardcoded secrets; configuration for transport type and port relies on environment variables. The HTTP server setup uses standard practices with Express. A full security audit of the underlying `@modelcontextprotocol/sdk` and `express` dependencies is beyond the scope of this source code analysis.
Updated: 2025-12-12GitHub→
0
0
Low Cost
BhumitArora icon

remote-MCP

by BhumitArora

Sec8

Manages personal or small-scale expense tracking, allowing users to add, list, and summarize financial transactions.

Setup Requirements

  • ⚠️Database stored in a temporary directory, meaning data will be lost upon server restart or temporary directory cleanup.
  • ⚠️Requires `fastmcp` and `aiosqlite` Python packages to be installed.
Verified SafeView Analysis
The server correctly uses parameterized queries to prevent SQL injection. No 'eval' or obfuscation is present, and no hardcoded secrets were found. The database is stored in a temporary directory, which can be a security benefit by limiting persistence but also means data loss. Error messages might expose minor database details, which is a minor information disclosure risk. The server listens on 0.0.0.0, which is standard but requires proper firewalling in production.
Updated: 2025-12-12GitHub→
0
0
Medium Cost
DevSkillsIT icon
Sec4

Enables AI assistants (Claude Code, Gemini CLI, Claude Desktop) to interact with Veeam Backup & Replication for monitoring, querying, and controlling backup infrastructure through natural language.

Setup Requirements

  • ⚠️Requires Veeam Backup & Replication 12+ with REST API enabled.
  • ⚠️Veeam API credentials with appropriate permissions (read-only for queries, higher for control operations) are mandatory.
  • ⚠️Node.js 20+ runtime environment is required.
  • ⚠️The `VEEAM_IGNORE_SSL=true` setting is enabled by default, posing a significant security risk by ignoring invalid SSL certificates for the Veeam API. This must be explicitly set to `false` and proper SSL configured for production deployments.
  • ⚠️A custom `AUTH_TOKEN` environment variable must be set by the user for securing the MCP HTTP endpoints.
Review RequiredView Analysis
The server's default configuration, especially in `docker-compose.yml` and explicitly in several tool files, sets `VEEAM_IGNORE_SSL` to `true` or `rejectUnauthorized` to `false`. This disables SSL certificate validation for connections to the Veeam REST API, making the communication vulnerable to Man-in-the-Middle (MITM) attacks. While `SafetyGuard` is implemented for critical write operations (`start-backup-job`, `stop-backup-job`) requiring explicit confirmation tokens and audit logging, and sensitive credentials are sourced from environment variables, the default insecure SSL configuration is a critical vulnerability for production environments. The HTTP server runs without native HTTPS, relying on external reverse proxy for SSL in production, which is a documented but insecure default for direct exposure.
Updated: 2025-12-20GitHub→
0
0
Low Cost
pyhoon icon
Sec8

A simple inventory management system demonstrating MCP Server functionality with SQLite through standard input/output.

Setup Requirements

  • ⚠️Requires Java Development Kit (JDK) installed and configured in the PATH or specified in the Claude Desktop config.
  • ⚠️Requires the compiled 'mcp_server.jar' file to be present at the exact path specified in the Claude Desktop configuration (e.g., C:\B4X\Development\mcp-server\Objects\mcp_server.jar).
Verified SafeView Analysis
The server runs locally via standard I/O, minimizing network exposure. No 'eval' or obvious malicious patterns are present in the truncated code. The primary security consideration would be potential SQL injection vulnerabilities within the unseen B4J implementation of 'get_inventory' and 'add_product' if user inputs are not properly sanitized before database interaction. Assuming standard safe coding practices in B4J, the risk is moderate.
Updated: 2025-12-12GitHub→
0
0
Medium Cost
FairArena icon
Sec9

Provides a Model Context Protocol (MCP) server for semantic search and retrieval of FairArena documentation using Pinecone and Google Gemini embeddings.

Setup Requirements

  • ⚠️Requires Pinecone API key and configured Pinecone indexes (developer and user documentation).
  • ⚠️Requires a Google Gemini API Key for effective semantic search; otherwise, it falls back to a hash-based (non-semantic) embedding for development purposes.
  • ⚠️The embedding dimension in environment variables must match the Pinecone index configuration.
Verified SafeView Analysis
The server implements strong security practices including API key authentication, rate limiting (per IP and per API key), secure headers (Helmet), and Zod-based request validation. Sensitive data is redacted in structured logs. Circuit breakers are used for external services to prevent cascading failures. The primary API key validation is based on format, with actual validation occurring when making calls to the Google API, which is a reasonable approach for third-party keys. CORS is configurable but defaults to '*' if not explicitly set, which should be adjusted for production deployments.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
pfaeffli icon
Sec9

MCP server wrapper for the Clockodo time tracking API, offering time tracking, HR analytics, and team management with role-based access.

Setup Requirements

  • ⚠️Requires Clockodo API credentials (CLOCKODO_API_USER, CLOCKODO_API_KEY).
  • ⚠️Requires Python 3.12+ (if running directly from source).
  • ⚠️Docker is recommended for simplified setup and local client integration.
  • ⚠️The HTTP/SSE transport (CLOCKODO_MCP_TRANSPORT=sse) is experimental, has known issues, and lacks built-in authentication, making it unsuitable for direct internet exposure without external security infrastructure.
Verified SafeView Analysis
The server follows good security practices: API credentials are exclusively loaded from environment variables and not hardcoded. Sensitive information is masked in logs. The codebase is structured with clear separation of concerns, and `httpx` is used with proper error handling. Multi-stage Docker builds reduce attack surface. The project explicitly warns about the experimental HTTP/SSE transport's lack of built-in authentication, rate limiting, and audit trails, recommending it only for development or behind a robust reverse proxy with external security measures. The core code itself is well-secured.
Updated: 2026-01-19GitHub→
PreviousPage 403 of 760Next