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
Viky11992 icon
Sec1

Unable to determine without source code for analysis. The repository name 'MCP-SERVER-CONNECT' suggests it might relate to connecting to a server, potentially for a game or specific protocol.

Setup Requirements

  • ⚠️No source code provided for analysis, therefore specific requirements or friction points cannot be identified.
Review RequiredView Analysis
No source code was provided for analysis. Therefore, a security audit cannot be performed, and safety cannot be guaranteed. It is critical to review any code before running.
Updated: 2025-11-26GitHub→
0
0
Medium Cost
Haervwe icon

timelines-mcp

by Haervwe

Sec9

Provides an MCP server for LLMs to manage and maintain coherent time-dependent narratives including timelines, events, and characters.

Setup Requirements

  • ⚠️Requires Python 3.12 or newer.
  • ⚠️PostgreSQL and Qdrant database adapters are 'not yet implemented' (raises NotImplementedError), limiting production database choices to SQLite and ChromaDB at this stage.
  • ⚠️For running setup/test scripts using `uv run`, `uv` must be installed globally (e.g., `pip install uv`).
Verified SafeView Analysis
The server implements robust JWT authentication for HTTP mode and gracefully handles its absence in STDIO mode for single-user scenarios. Database interactions use parameterized queries, preventing SQL injection. While local development scripts use a placeholder secret key, the documentation clearly advises against using it in production and recommends asymmetric key authentication (JWKS) for enhanced security. No obvious code injection vulnerabilities were found in the server's core logic.
Updated: 2025-12-05GitHub→
0
0
Low Cost
mew-ton icon

kex

by mew-ton

Sec8

Kex functions as a librarian tool for AI agents and humans, providing keyword-indexed coding guidelines and documentation to AI agents on demand to optimize context and reduce token usage during development.

Setup Requirements

  • ⚠️Requires Go 1.25+ to be installed for installation via `go install`.
  • ⚠️Requires a compatible AI agent/editor (e.g., Claude Desktop/VSCode, Claude Code) with MCP client setup to fully utilize its features.
  • ⚠️If using remote references or the central repository use case, it depends on a static site host (e.g., GitHub Pages) for serving generated content.
Verified SafeView Analysis
The application serves markdown content over an MCP (Model Context Protocol) server. While the Go code itself does not directly perform dangerous operations like 'eval' on arbitrary user input, it fetches and serves markdown documents from local files and potentially remote URLs. If a remote source (e.g., a GitHub Pages site) were compromised or untrusted, it could serve malicious markdown. The responsibility of sanitizing and safely rendering this markdown content falls on the connected AI client/editor, not Kex itself. There are no hardcoded secrets visible in the provided truncated source code, and `os/exec` is only observed in e2e tests, not core server runtime. The `KEX_REMOTE_TOKEN` environment variable or `.kex.yaml` configuration is used for authenticated remote fetches.
Updated: 2026-01-18GitHub→
0
0
Low Cost
Genusboragosirharoldwalterkroto654 icon

claude-config-editor

by Genusboragosirharoldwalterkroto654

Sec9

A local web-based GUI tool to clean, optimize, and manage Claude AI configuration files (e.g., chat histories, project sizes, server configurations).

Setup Requirements

  • ⚠️Requires Python 3 to run.
  • ⚠️Requires an existing Claude Code or Claude Desktop configuration file to be useful.
Verified SafeView Analysis
The server runs locally on `localhost:8765` and does not expose itself to the public internet by default. It primarily reads from and writes to user-specific Claude configuration files within the home directory, which is its stated purpose. The code is straightforward, uses Python's standard library, and includes an auto-backup mechanism before saving, reducing the risk of data loss. No `eval` or obfuscation found, and no hardcoded secrets. The main risk is inherent to any tool modifying user files, but the design choices minimize this risk.
Updated: 2026-01-19GitHub→
0
0
High Cost
ToxicFartCloud icon

gemini-cli-artilect

by ToxicFartCloud

Sec10

Transforms the Gemini CLI into an expert Software Architect, enforcing rigorous engineering standards and providing deep technical guidance based on a curated knowledge library.

Setup Requirements

  • ⚠️Requires Node.js installed to run the local knowledge base server.
Verified SafeView Analysis
The provided source code defines an AI persona and knowledge base content. There are no executable scripts or direct network risks within the provided files. The extension's MCP server uses a standard, trusted filesystem server. The persona itself explicitly advocates for 'Bastion Security' protocols and avoiding hardcoded secrets, indicating a security-conscious design philosophy.
Updated: 2025-11-29GitHub→
0
0
Medium Cost
BrianLusina icon
Sec3

An image research assistant that uses modular services for vision and Wikipedia lookups to answer user queries.

Setup Requirements

  • ⚠️Requires `GOOGLE_GEMINI_API_KEY` (a paid service) to function.
  • ⚠️Requires Python 3.13 or newer (as per `pyproject.toml`).
  • ⚠️The Gradio web UI launches on `0.0.0.0` and should be run in a controlled environment or behind a proxy for production to mitigate exposure.
Review RequiredView Analysis
The `load_image_from_path` tool directly uses an unsanitized `file_path` argument to open files. If the agent is susceptible to prompt injection, a malicious user could potentially trick the agent into using this tool to read arbitrary files from the server's filesystem (Local File Inclusion/Disclosure). The Gradio UI exposes a web interface, which, if not properly secured, could make this vulnerability exploitable. While API keys are handled via environment variables (good practice), this file access vulnerability is critical.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
nhatvu148 icon
Sec8

Provides an MCP server to transcribe videos from URLs (1000+ platforms) or local files using OpenAI's Whisper model, generating transcripts in TXT, JSON, and Markdown formats.

Setup Requirements

  • ⚠️Requires external system dependencies (yt-dlp, openai-whisper, ffmpeg) to be manually installed on the host system.
  • ⚠️Installation instructions for dependencies are platform-specific (macOS, Linux, Windows) and involve various package managers (brew, apt, pip, winget, choco).
  • ⚠️The 'openai-whisper' dependency necessitates a Python installation with 'pip' available on the system.
Verified SafeView Analysis
The server executes external command-line tools (yt-dlp, whisper, ffmpeg) via `child_process.exec`. While input arguments are generally quoted to mitigate shell injection, the overall security relies on the integrity of these external binaries and the host system. A large `maxBuffer` for `child_process.exec` (10MB) could be a minor denial-of-service vector if large outputs are maliciously crafted. No `eval` or hardcoded secrets found.
Updated: 2025-11-26GitHub→
0
0
Low Cost
quickitquote icon

qiq-mcp-server

by quickitquote

Sec1

A custom JSON-RPC 2.0 WebSocket/HTTP server designed for OpenAI Agent Builder, providing specialized tools like Typesense search and QIQ scoring.

Setup Requirements

  • ⚠️Requires `MCP_TOKEN` for server authentication.
  • ⚠️Requires a Typesense account and associated API keys (`TYPESENSE_HOST`, `TYPESENSE_SEARCH_ONLY_KEY`, `TYPESENSE_COLLECTION`, etc.) if the `typesense_search` tool is enabled.
  • ⚠️Requires Node.js (version 18+ recommended by project dependencies).
Review RequiredView Analysis
Production API keys and access tokens (`MCP_TOKEN`, `TYPESENSE_SEARCH_ONLY_KEY`) are explicitly listed with live values in the `Docs/MCP-Deployment-Guide.md` and `Docs/GitHub-Secrets-Setup.md` files, which are part of the repository. This constitutes a severe security vulnerability as these secrets are publicly exposed. Additionally, the server supports token authentication via URL query parameters (`?token=`), which is generally less secure than header-based authentication as tokens in URLs can be exposed in server logs, browser history, or referer headers.
Updated: 2025-12-15GitHub→
0
0
Low Cost
itsalaz icon

mcp-server

by itsalaz

Sec9

This server uses the Model Context Protocol (MCP) SDK to define and expose two tools and two static resources, primarily for fetching mock weather data and tracking packages, communicating via standard I/O.

Setup Requirements

  • ⚠️Requires Node.js runtime (v18+ recommended)
  • ⚠️Requires TypeScript compilation before execution (e.g., `tsc server.ts` to generate `server.js`)
Verified SafeView Analysis
The code is straightforward with no obvious malicious patterns or usage of dangerous functions like 'eval'. Input validation is handled using Zod. The server uses StdioServerTransport, meaning it communicates via standard I/O, which inherently limits direct network exposure, reducing certain external attack vectors. Mock data is returned by tools, preventing external data injection in those specific functions. The only potential minor risk is JSON.stringify on potentially untrusted data, but in this specific mock context, it's not a direct vulnerability.
Updated: 2025-11-19GitHub→
0
0
Low Cost
Gnanesh09 icon
Sec4

This server provides authenticated API endpoints for a 'Notes app' using the FastMCP framework, leveraging Stytch for bearer token authentication.

Setup Requirements

  • ⚠️Requires `STYTCH_DOMAIN` and `STYTCH_PROJECT_ID` environment variables to be configured for authentication to function.
  • ⚠️Requires Python 3.11 or newer.
Verified SafeView Analysis
CORS is configured with `allow_origins=["*"]`, `allow_credentials=True`, `allow_methods=["*"]`, and `allow_headers=["*"]`. While this might be convenient for local development, it is highly insecure for production environments and creates a critical vulnerability if deployed as-is, as it allows any origin to make credentialed requests. The use of environment variables for Stytch credentials is a good practice.
Updated: 2025-11-30GitHub→
0
0
High Cost
klogdog icon

EasyMCP

by klogdog

Sec4

A Docker-native server factory that automatically builds, configures, and deploys Model Context Protocol (MCP) servers from user-defined tools and connectors in TypeScript or Python.

Setup Requirements

  • ⚠️Requires Docker 20.10+ (with `--privileged` access to host Docker daemon for generator)
  • ⚠️Requires Node.js 20+ and npm 9+
  • ⚠️Requires Python 3.8+ if using Python-based tools or connectors
  • ⚠️OpenAI API Key (Paid) is often required for advanced AI/ML tools/connectors like translation or classification beyond basic mock functionality.
Review RequiredView Analysis
The MCP generator itself requires `--privileged` Docker access and mounts `/var/run/docker.sock` to the host. This grants the container full root access to the host's Docker daemon, making it highly sensitive. If the generator or its inputs (tools/connectors) are compromised, it could lead to host system compromise. While this is intentional for its Docker-in-Docker build functionality, it necessitates extreme trust in the generator's code and all dynamic modules it processes. Dynamic execution of Python tools via `child_process` also presents a potential risk if inputs are not thoroughly sanitized. The project uses `zod` for schema validation and a `SecretManager` for credential handling, which are good practices, but the fundamental privilege requirement remains a critical security consideration.
Updated: 2025-11-27GitHub→
0
0
Low Cost
ohadbenami icon

supabase-mcp-server

by ohadbenami

Sec3

Provides a Model Context Protocol (MCP) server to integrate Supabase databases with Claude, enabling CRUD operations, custom SQL queries, and real-time data management.

Setup Requirements

  • ⚠️Requires Node.js 18+.
  • ⚠️Requires a Supabase account and project.
  • ⚠️Requires `SUPABASE_URL` and `SUPABASE_API_KEY` to be set as environment variables.
Review RequiredView Analysis
The `execute_sql` tool allows passing an arbitrary SQL query string. While it relies on a Supabase RPC function (which might be secured on the database side), exposing direct SQL execution to an LLM via an unvalidated string input is a critical security risk for SQL injection and potential full database compromise if the API key used has elevated privileges. The note 'read-only recommended' is not a technical enforcement.
Updated: 2025-11-27GitHub→
PreviousPage 321 of 713Next