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
High Cost
BACH-AI-Tools icon

chrome-devtools-mcp

by BACH-AI-Tools

Sec4

Enables AI coding agents to control and inspect a live Chrome browser for automation, debugging, and performance analysis.

Setup Requirements

  • ⚠️Requires Node.js v20.19 or newer LTS, and a current stable Chrome browser or newer.
  • ⚠️Operating system sandboxes (e.g., macOS Seatbelt, Linux containers) may prevent the server from launching Chrome; requires disabling sandboxing or connecting to a manually started Chrome instance.
  • ⚠️Connecting to a running Chrome instance requires specific Chrome launch flags (--remote-debugging-port, --user-data-dir) and manual setup.
Review RequiredView Analysis
The `evaluate_script` tool allows the AI agent to execute arbitrary JavaScript code within the browser context. While this is its intended functionality, it poses a significant security risk if the AI agent or the client communicating with it is compromised, as it could lead to arbitrary code execution on visited web pages. Additionally, `--proxyServer` and `--acceptInsecureCerts` options, if misused, can expose traffic or make connections vulnerable. The project explicitly warns users about exposing sensitive information to MCP clients.
Updated: 2026-01-04GitHub→
0
0
Medium Cost
FurkanEmiroglu icon

Applovin-Max-MCP-Server

by FurkanEmiroglu

Sec3

Provides LLM clients like Claude with access to AppLovin Max advertising platform analytics and reporting capabilities for mobile ad revenue data.

Setup Requirements

  • ⚠️Requires Go 1.25.4 or later
  • ⚠️Requires an AppLovin Max account with API access
  • ⚠️Requires setting APPLOVIN_API_KEY environment variable, which is then transmitted in API query parameters
Review RequiredView Analysis
Explicitly labeled as 'PROOF OF CONCEPT ONLY' and 'NOT production-ready' with a detailed list of known limitations including: no comprehensive error handling, no rate limiting, no input validation beyond basic checks, no logging, no tests, no security audits, and critically, API keys are transmitted in query parameters. While transmitted over HTTPS, this is a less secure method than header authentication and is prone to logging by proxies or web servers. Use with extreme caution and not in production.
Updated: 2025-12-02GitHub→
0
0
Medium Cost
o98k-ok icon

meme

by o98k-ok

Sec2

A high-performance emoji search service that aggregates results from multiple sources and provides them to AI clients (e.g., Claude Desktop, Cursor) via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires users to set up their own image proxy service and configure `IMAGE_PROXY_URL` for sources that have anti-hotlinking measures (e.g., Qudoutu, Doutub).
  • ⚠️Requires manual extraction and configuration of `DOUYIN_COOKIE` from a browser for searching Douyin memes.
Review RequiredView Analysis
CRITICAL: The HTTP client used for web scraping in `internal/sources/scrapers.go` and `internal/sources/api.go` explicitly sets `InsecureSkipVerify: true` within its `tls.Config`. This disables TLS certificate verification, making all HTTPS connections vulnerable to Man-in-the-Middle (MITM) attacks. This allows potential interception and tampering of sensitive data (like `DOUYIN_COOKIE` or search queries) and fetched images without detection, posing a severe security risk.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
madebymlai icon

spec-context-mcp

by madebymlai

Sec8

Unified MCP server combining semantic code search with spec-driven development workflow, providing a dashboard for managing specs, approvals, and implementation logs, and supporting AI-assisted features.

Setup Requirements

  • ⚠️Requires OPENROUTER_API_KEY for LLM integrations (paid service).
  • ⚠️Requires a running Qdrant vector database instance (e.g., via Docker: `docker run -p 6333:6333 qdrant/qdrant`).
  • ⚠️Requires Node.js runtime environment.
Verified SafeView Analysis
API keys (OpenRouter, Qdrant) are handled via environment variables and not hardcoded. Path handling in the backend uses `PathUtils` for normalization and validation, mitigating directory traversal risks. The dashboard frontend sanitizes user-provided content before rendering with `dangerouslySetInnerHTML` to prevent common XSS attacks. Rate limiting and audit logging are implemented in the dashboard's `security-utils.ts`. A theoretical, minor XSS risk exists with Mermaid diagram rendering (`securityLevel: 'loose'`) when combined with `dangerouslySetInnerHTML` if malicious diagram code is processed, although content is typically internal and trusted within the development workflow context.
Updated: 2026-01-18GitHub→
0
0
Low Cost
Ryno-Crypto-Mining-Services icon

braiins-insights-mcp-server

by Ryno-Crypto-Mining-Services

Sec9

A Model Context Protocol (MCP) server exposing Braiins Insights Dashboard API for Bitcoin network analytics, mining statistics, and profitability metrics to AI coding assistants.

Setup Requirements

  • ⚠️Node.js 18+ and TypeScript 5.0+ are required.
  • ⚠️Manual configuration of an MCP client (e.g., Claude Desktop, Cursor IDE) is necessary to use the server.
  • ⚠️Client-side rate limiting may issue 'Network Error' responses if the tool is called too frequently (30 req/min, 5 req/sec burst), even if the upstream API doesn't explicitly rate limit.
Verified SafeView Analysis
The server is explicitly designed to interact with a public API (Braiins Insights Dashboard API) that requires no authentication, eliminating API key leakage risks for the Insights API. Client-side rate limiting and Zod-based input validation are implemented to prevent abuse and ensure data integrity. No hardcoded secrets are present in the server's runtime code. GitHub secrets mentioned (NPM_TOKEN, CODECOV_TOKEN, SNYK_TOKEN) are for CI/CD pipeline operations, not for the server's runtime.
Updated: 2026-01-19GitHub→
0
0
Low Cost
Cronos402 icon

app

by Cronos402

Sec4

Main web application and payment gateway for Cronos Model Context Protocol (MCP) services, enabling server registration, wallet authentication, payment processing, and usage analytics.

Setup Requirements

  • ⚠️Requires a PostgreSQL database.
  • ⚠️Depends on several other Cronos402 microservices (Auth, MCP2, MCP Data, API2, Facilitator) running on specific ports, implying a complex distributed setup.
  • ⚠️The `NEXT_PUBLIC_WALLETCONNECT_PROJECT_ID` environment variable is critical for production wallet connectivity and defaults to a 'demo-project-id'.
Review RequiredView Analysis
The `mcp-proxy` API route (`src/app/api/mcp-proxy/route.ts`) is vulnerable to Server-Side Request Forgery (SSRF). It decodes a `target-url` parameter from the request and uses it directly in a server-side `fetch` call without sufficient URL validation or whitelisting. This allows an authenticated user to potentially make arbitrary requests from the server to internal network resources or external hosts. While the application uses Drizzle ORM for database interactions and `better-auth` for authentication, this specific proxy vulnerability is critical. There are no explicit hardcoded secrets observed, and text sanitization is used for display purposes, which are good practices. The project mentions 'TODO: add withProxy and LoggingHook back in' which suggests intended security/logging features are currently absent from the proxy.
Updated: 2026-01-19GitHub→
0
0
Medium Cost

error-logger-mcp

by zetaphoenix888-byte

Sec5

An MCP server for logging errors/solutions and managing behavioral rules for Claude Code development, enabling knowledge sharing and best practice enforcement across development environments.

Setup Requirements

  • ⚠️Requires Node.js >= 18.0.0.
  • ⚠️TypeScript compilation (`npm run build`) is mandatory before running the server directly (e.g., use `npm run start:safe` or `npm run server:safe`).
  • ⚠️For remote (LAN) access, ensure `ERROR_LOGGER_PIN` is set in the `.env` file or environment variables if `ERROR_LOGGER_REQUIRE_AUTH=true` is desired, as a missing PIN will disable authentication.
Verified SafeView Analysis
The server's authentication for remote (LAN) access relies on an optional 4-digit PIN. A critical vulnerability exists where if `ERROR_LOGGER_REQUIRE_AUTH` is set to `true` but `ERROR_LOGGER_PIN` is not configured, remote authentication is *disabled* instead of access being denied. This means remote access might be unintentionally open. Localhost access is exempt. CORS is restricted to local network IPs, limiting exposure. File I/O operations utilize `proper-lockfile` for concurrency control. Input validation includes checks for encoding corruption patterns to prevent data integrity issues, but extensive data validation (e.g., specific string length limits beyond basic checks for all fields) is noted as limited in the documentation.
Updated: 2025-11-27GitHub→
0
0
Low Cost
hegner123 icon
Sec9

Coordinates multiple AI agents for parallel development tasks by managing atomic step claiming, dependency tracking, and crash recovery using a SQLite database.

Setup Requirements

  • ⚠️Requires Go 1.21+ to build and run.
  • ⚠️Requires SQLite 3.35+ for WAL mode functionality.
  • ⚠️Database is a local SQLite file, not suitable for multi-machine distributed coordination without external solutions.
Verified SafeView Analysis
The server design explicitly addresses security concerns for its intended local execution context. It uses parameterized queries to prevent SQL injection, transports data via stdio (no network exposure), and relies on file permissions for database access. There are no hardcoded secrets or 'eval' patterns identified in the provided source code. Agents are trusted as they operate under the same user's context. The database file is stored in the user's home directory (`~/.claude/agent-coordination.db`) which is reasonable for a local tool.
Updated: 2026-01-18GitHub→
0
0
Medium Cost
hockinghills icon

memento-cloudflare

by hockinghills

Sec4

A Cloudflare Worker acting as a Model Context Protocol (MCP) server, exposing a Neo4j knowledge graph through hybrid semantic and keyword search capabilities, powered by VoyageAI embeddings.

Setup Requirements

  • ⚠️Requires a Neo4j database instance (e.g., Neo4j Aura) with its HTTP Query API enabled.
  • ⚠️Requires a VoyageAI API Key, which is for a paid embedding service.
  • ⚠️Requires deployment to Cloudflare Workers and a configured Cloudflare KV Namespace for OAuth (if expanded beyond current static metadata endpoint).
Verified SafeView Analysis
The `test-vector-http.js` file contains hardcoded `VOYAGE_API_KEY` and `NEO4J_AUTH` credentials, which is a critical security vulnerability as it exposes sensitive production-like secrets directly in the codebase. While the main worker (`src/index.ts`) is configured to use environment variables (a good practice), the presence of these secrets in the repository significantly lowers its security posture. The worker's current `fetch` handler only returns static metadata, making it safe as deployed. However, the comments indicate future plans to expose the full tool capabilities, which would require robust OAuth (mentioned as 'next') to prevent unauthorized database access and API usage. Cypher queries are parameterized, mitigating direct SQL/Cypher injection.
Updated: 2025-12-19GitHub→
0
0
Medium Cost
klietus icon

SignalZeroMCP

by klietus

Sec9

The server provides a Meta-Controller Protocol (MCP) interface for a SignalZero shared symbolic store, wrapping an AWS-based symbol store API.

Setup Requirements

  • ⚠️Requires an AWS account and proper deployment of the SignalZero backend API (via CloudFormation or existing infrastructure) if not using the default public endpoint.
  • ⚠️Requires an API Key for the backend SignalZero API to be set as the `SYMBOL_STORE_API_KEY` environment variable.
  • ⚠️Python dependencies must be installed via `pip install -r requirements.txt`.
Verified SafeView Analysis
The server design is robust, acting as a proxy to an external AWS API. It correctly uses environment variables for sensitive data like API keys. There are no obvious 'eval' or similar dangerous patterns. Input parsing for `PUT` requests is confined to JSON, and query parameters are handled safely. The underlying AWS CloudFormation template indicates secure practices such as API key requirements and IAM roles with least privilege. The primary risk would be misconfiguration of the AWS backend or vulnerabilities in external dependencies.
Updated: 2025-11-27GitHub→
0
0
Low Cost
MatrixRonny icon

FileMcpServer

by MatrixRonny

Sec1

A server application potentially involved in file management or resource control, based on its name.

Review RequiredView Analysis
Source code was not provided in the prompt. Therefore, a comprehensive security audit is impossible, and the project's safety cannot be verified. The extremely low score reflects the critical unknown risks due to the lack of code access.
Updated: 2025-12-01GitHub→
0
0
Medium Cost
jayakumarpujar icon
Sec4

This server is designed to integrate an AI agent built with Google's ADK and Gemini 2.0 Flash with external data services via the Airbnb Model Context Protocol (MCP) to enable autonomous LLM interactions.

Setup Requirements

  • ⚠️Requires Google Agent Development Kit (ADK) knowledge and setup.
  • ⚠️Requires Google Cloud API access for Gemini 2.0 Flash (paid service).
  • ⚠️Configuration and integration with various external data services will be necessary.
Review RequiredView Analysis
The provided source code is extremely truncated, containing only the README.md. Critical security aspects such as 'eval' usage, obfuscation, network risks, hardcoded secrets, or malicious patterns cannot be assessed without access to the actual code files. Therefore, the safety cannot be verified.
Updated: 2026-01-16GitHub→
PreviousPage 336 of 713Next