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
bugzy-ai icon

jira-mcp-server

by bugzy-ai

Sec9

Enables AI assistants to interact with on-premise Jira Server deployments for issue management.

Setup Requirements

  • ⚠️Requires Node.js 20.0.0 or higher.
  • ⚠️Jira Server/Data Center 8.14+ is required for Personal Access Token (PAT) authentication; older versions necessitate less secure Basic Auth.
  • ⚠️Local development setup with 'npm run setup' requires Docker and manual steps for Jira license and administrator account creation on first run.
Verified SafeView Analysis
The server uses environment variables for sensitive credentials (PAT or Basic Auth) and performs input validation with Zod. It logs sensitive request/response data to a separate file only when a debug flag is enabled, preventing exposure via standard output. The use of Basic Authentication for older Jira versions is inherently less secure than PAT, but is a documented functional requirement and not a vulnerability in the server itself.
Updated: 2025-11-25GitHub→
0
0
High Cost
Scarmonit icon
Sec2

Provides a multi-model AI prompt templating engine with variable substitution, orchestration, workflow automation, and cost optimization for various LLMs like Claude, GPT-4, Gemini, and Kimi.

Setup Requirements

  • ⚠️Requires API keys for Claude, GPT-4, Gemini, and Kimi (paid services).
  • ⚠️Requires YAML templates to be defined for prompt templating.
  • ⚠️Integration with an MCP server implies an external dependency or specific deployment environment.
Review RequiredView Analysis
Cannot perform a comprehensive security audit as only the README.md file was provided for 'SOURCE CODE'. No executable code, dependencies, or configuration files were available to check for 'eval', obfuscation, network risks, hardcoded secrets, or malicious patterns. The security_score reflects this critical lack of visibility into the actual codebase.
Updated: 2025-12-03GitHub→
0
0
Low Cost
Vitruveo icon

vtru-mcp

by Vitruveo

Sec9

Provides read-only blockchain services for the Vitruveo network to AI agents via the Model Context Protocol.

Setup Requirements

  • ⚠️Requires Bun 1.0+ or Node 18+
  • ⚠️Hardcoded to Vitruveo Mainnet (Chain ID 1490)
Verified SafeView Analysis
The server's source code contains functions for performing write operations (transfers, approvals, contract writes) that require a private key (`src/mcp/services/transfer.ts`). However, the `src/mcp/tools.ts` file explicitly comments out the registration of all such 'write' tools, aligning with the README's claim of being 'read-only' and 'safe for public use' with 'no private keys, no transfers or approvals, no write methods'. This means, as provided and configured, it adheres to its read-only guarantee. The presence of the underlying write functionality (even if commented out) is a minor concern as it could be enabled by a malicious actor or misconfiguration.
Updated: 2025-11-19GitHub→
0
0
Low Cost
GHjiejie icon
Sec7

Provides a web-based API and frontend for a control program or service.

Setup Requirements

  • ⚠️Requires Node.js to be installed.
  • ⚠️A frontend build step (`npm run build`) is necessary to generate the static assets (`dist` folder) that the Node.js Express server needs to serve. The server's `src/index.js` currently points to a `public` directory, while `webpack.config.js` outputs to `dist`, indicating a potential path mismatch that would need to be resolved for the application to function correctly.
  • ⚠️The `start` script in `package.json` only runs a `webpack-dev-server` for frontend development, not the Node.js backend server.
Verified SafeView Analysis
The provided source code is heavily truncated, making a comprehensive security audit impossible. Based on the visible `src/index.js` and configuration files, there are no explicit 'eval' calls, direct obfuscation, or hardcoded secrets. The application is an Express server serving static files, which introduces standard web security considerations (e.g., XSS, CSRF, path traversal) that cannot be fully assessed without the complete application code and a deeper understanding of its functionality. However, no immediate critical vulnerabilities are apparent in the provided snippets.
Updated: 2025-12-03GitHub→
0
0
High Cost
i11v icon
Sec8

Provides AI assistants with Model Context Protocol (MCP) access to Fastmail accounts for email management.

Setup Requirements

  • ⚠️Requires a Fastmail API Token with appropriate scopes (Mail read/write) to be generated manually.
  • ⚠️Requires configuring Claude Code's MCP settings with the server URL and API token in `~/.claude/claude_desktop_config.json` or `.mcp.json`.
  • ⚠️Requires `REDIS_URL` environment variable for caching, especially in production deployments like Vercel.
  • ⚠️Uses `pnpm` as the package manager, not `npm` or `yarn`.
Verified SafeView Analysis
The server correctly extracts the Fastmail API token from the Authorization header. It hashes this token before using it as a Redis key, avoiding storage of raw tokens in Redis keys. However, the full session JSON, which can contain sensitive account IDs, is cached in Redis. While given a TTL, a compromise of the Redis instance could expose this data. Input validation for tools is enforced using Zod schemas. The server explicitly warns users to keep their API tokens secure and not commit them to version control. The open CORS configuration (origin: '*') is typical for MCP but means no origin-based protection beyond the bearer token.
Updated: 2026-01-18GitHub→
0
0
Low Cost
ashleyappadoo icon

timesfm-mcp

by ashleyappadoo

Sec6

Optimized time series forecasting server for cloud platforms like Railway/Render, offering trend and seasonality analysis via an HTTP API.

Setup Requirements

  • ⚠️The provided `timesfm_server.py` implements a custom 'pure Python' forecasting algorithm (seasonal trend decomposition + linear regression) and does not use Google Research's TimesFM model, despite the repository name and README claiming 'Vrai TimesFM 1.0'. This means the forecasting quality and methodology will differ significantly from actual TimesFM.
  • ⚠️Deployment to Railway/Render, as described in the README, requires manually replacing three specific files (`Dockerfile`, `requirements.txt`, `timesfm_server.py`) in the user's GitHub repository.
  • ⚠️The forecast horizon is capped at a maximum of 32 periods internally by the `timesfm_server.py`.
Verified SafeView Analysis
The server is implemented in pure Python using `http.server`, which is not typically recommended for production environments due to limitations in robustness, performance, and advanced security features (like TLS, advanced input validation, or DDoS protection) compared to WSGI servers. It lacks explicit authentication or authorization, meaning the API is open to anyone with the URL. Request logging from the base HTTP server is suppressed, though application-level logs exist. No malicious patterns, `eval` usage, or hardcoded secrets were identified.
Updated: 2026-01-16GitHub→
0
0
Low Cost

MCP_Server-Repo

by praveen1210

Sec1

The use case for this server cannot be determined as the repository's README file was not found (404: Not Found).

Review RequiredView Analysis
Cannot perform a security audit due to the complete lack of accessible repository content (README 404: Not Found). Assuming lowest safety score as no information allows for assessment.
Updated: 2025-11-18GitHub→
0
0
Low Cost
spilchen icon
Sec9

An MCP (Model Context Protocol) server that provides AI assistants and other clients read-only access to Yahoo Fantasy Sports data.

Setup Requirements

  • ⚠️Requires valid Yahoo API credentials (either `oauth2.json` file or `YAHOO_CLIENT_ID` and `YAHOO_CLIENT_SECRET` environment variables). Registering an application with Yahoo Developer Network is necessary.
  • ⚠️Requires `YAHOO_LEAGUE_ID` environment variable set to a valid Yahoo Fantasy league ID. The server provides a helper to list available league IDs if not set.
  • ⚠️If using environment variables for credentials, an initial manual OAuth flow via browser is required to obtain tokens.
Verified SafeView Analysis
The server explicitly states it provides read-only access, significantly reducing risk. Authentication details (consumer keys, secrets, tokens) are handled via environment variables or a local `oauth2.json` file, which are standard and secure practices, avoiding hardcoding. The `stdio_transport` uses `pydantic.BaseModel.model_validate_json` for incoming JSON-RPC messages, which is robust against typical JSON parsing vulnerabilities. No `eval` or other highly dangerous functions are directly used for dynamic code execution. The main risks would be in the underlying `yahoo_fantasy_api` library or the Yahoo API itself.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
ermermermermidk icon

mcp-ai-memory

by ermermermermidk

Sec9

This server provides a Model Context Protocol (MCP) interface for managing an AI's semantic memory, enabling storage, retrieval, clustering, and consolidation of contextual knowledge.

Setup Requirements

  • ⚠️Requires PostgreSQL with the `pgvector` extension enabled.
  • ⚠️Requires Node.js 20+ or Bun runtime environment.
  • ⚠️Optional: Redis is highly recommended for caching and asynchronous job queuing (e.g., embedding generation, batch processing, clustering).
Verified SafeView Analysis
The server demonstrates strong security practices, including comprehensive input validation and sanitization using Zod schemas for all API inputs, which helps prevent common vulnerabilities like injection attacks. It uses Kysely ORM for database interactions, providing built-in protection against SQL injection. Sensitive configurations like database and Redis URLs are loaded from environment variables (dotenv), avoiding hardcoded secrets. There's no observable direct use of `eval` or unsafe `child_process` calls. The `SECURITY.md` outlines best practices for deployment (e.g., strong credentials, SSL/TLS for DB/Redis), acknowledging that external infrastructure security is critical.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
Gooseus icon

mcp-nats

by Gooseus

Sec9

Exposes NATS messaging capabilities (publish, subscribe, JetStream, KV, Object Store) to LLMs, enabling AI assistants to interact with real-time and persistent data infrastructure.

Setup Requirements

  • ⚠️Requires a NATS server with JetStream enabled.
  • ⚠️Requires Node.js 20+ (or Bun for development).
Verified SafeView Analysis
The server uses Zod for input validation on all tool parameters, which significantly reduces the risk of injection attacks or unexpected input. NATS authentication details are handled via environment variables or a credentials file, avoiding hardcoded secrets. File system access for embedded documentation is restricted to a specific `docs/nats` directory, preventing arbitrary file reads. The SSE transport sets permissive CORS headers (`Access-Control-Allow-Origin: *`), which is common for tools intended for LLM integration but should be considered in a broader network deployment context.
Updated: 2025-12-16GitHub→
0
0
High Cost

dJEB_mcp_server

by dPhoeniixx

Sec7

Bridges Claude Desktop with JEB decompiler for AI-assisted Android APK reverse engineering workflows, enabling advanced analysis directly within Claude.

Setup Requirements

  • ⚠️Requires a licensed version of JEB Decompiler.
  • ⚠️Requires Claude Desktop to be installed.
  • ⚠️The Claude Desktop configuration requires an absolute path to `jeb_mcp_bridge.py`.
Verified SafeView Analysis
The system involves running a script within JEB that exposes its API over a local socket to Claude Desktop via a Python bridge. While the connection is local (port 8851), exposing a powerful decompiler API to an AI client means that malicious or unexpected prompts could potentially trigger unintended actions through the JEB API. No explicit 'eval' is mentioned, but the nature of directly invoking decompiler functionality based on AI instructions requires trust in both the AI's understanding and the tool's input sanitization. Generally safe when used as intended by a knowledgeable user.
Updated: 2025-11-17GitHub→
0
0
Low Cost
mbocsi icon

gohab

by mbocsi

Sec3

A home automation server for managing IoT devices and sensors using a message-based architecture.

Setup Requirements

  • ⚠️Requires Go toolchain for building from source (uses `make` commands).
  • ⚠️LoRa transport (if enabled) requires a concrete `server.HardwareInterface` implementation for physical radio communication, as the provided `SX1276Radio` uses a mock interface.
  • ⚠️Default WebSocket transport allows all origins (`CheckOrigin: true`) which is a security risk in production environments.
Review RequiredView Analysis
Critical security vulnerabilities exist. The server lacks explicit authentication and authorization mechanisms for client (device) connections, allowing any client to connect and declare capabilities. The WebSocket transport `websocket.Upgrader` explicitly sets `CheckOrigin: true`, which disables same-origin policy enforcement and makes it vulnerable to Cross-Site WebSocket Hijacking (CSWH) if not mitigated by external security. This combination makes the system highly insecure for deployment in any exposed or production environment without significant security enhancements.
Updated: 2025-12-27GitHub→
PreviousPage 386 of 760Next