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
SackeyDavid icon

cursor-eod-mcp

by SackeyDavid

Sec7

An MCP server that automates the generation and posting of end-of-day (EOD) status updates to Slack, including auto-summarization of Cursor workspace conversations.

Setup Requirements

  • ⚠️Requires Node.js 18 or newer.
  • ⚠️Requires a Slack Workspace and a manually created Slack app with specific OAuth scopes (chat:write, channels:read, groups:read) to obtain Slack API tokens (xoxb- or xoxp-).
  • ⚠️Requires manual configuration in your IDE's MCP settings (~/.cursor/mcp.json or ~/.vscode/mcp.json) to specify the server command and Slack tokens (as environment variables or via the 'configure' tool).
Verified SafeView Analysis
Slack tokens are stored locally in an unencrypted JSON file (data/eod-mcp.json) if configured via the 'configure' tool, or expected as environment variables. This relies on file system permissions for security, which is a common practice for local tools but presents a vulnerability if the file system is compromised. No 'eval' or malicious code patterns were found. The use of `child_process.spawn` is limited to test files and does not pose a production risk.
Updated: 2025-12-05GitHub→
0
0
Low Cost
sritajkumarpatel icon

sritaj_info_mcp_server

by sritajkumarpatel

Sec10

This server provides static personal information about 'Sritaj' via the Model Context Protocol (MCP).

Setup Requirements

  • ⚠️Requires Python 3.13+
  • ⚠️Recommended use of 'uv' package manager
  • ⚠️Manual VS Code MCP configuration required for integration
Verified SafeView Analysis
The server's source code is simple, returning only static, hardcoded information. There are no dynamic code execution functions (like eval), no hardcoded secrets, and no external dependencies beyond the MCP framework. It poses minimal security risk.
Updated: 2025-12-03GitHub→
0
0
High Cost
Friday-AI-Core icon

adaptive-mcp

by Friday-AI-Core

Sec5

Integrates FridayAI’s ACW platform with any MCP-compatible environment for routing AI model calls, dynamic tool handling, and adaptive model switching.

Setup Requirements

  • ⚠️Requires access to FridayAI’s ACW platform
Review RequiredView Analysis
Source code for security analysis is limited to README.md. Cannot verify presence of 'eval', hardcoded secrets, network vulnerabilities, or malicious patterns. The server's role as an AI model router implies handling potentially sensitive data, making robust security implementation critical, but unassessable from the provided information.
Updated: 2025-11-23GitHub→
0
0
Medium Cost
schakrab02 icon

aiagent-st-fsi

by schakrab02

Sec7

A financial investment chatbot for stock analysis, providing information, research, analysis, recommendations, and human-in-the-loop trading order placement using Google ADK and Yahoo Finance data.

Setup Requirements

  • ⚠️Requires Google Cloud Project and Location setup.
  • ⚠️Requires a Google API Key for LLM access.
  • ⚠️Requires Python 3.12 or higher.
  • ⚠️Requires an external Yahoo Finance MCP server to be running and accessible (e.g., via `npx @modelcontextprotocol/inspector yahoo-finance-server`).
  • ⚠️Requires the Remote A2A Earnings Agent (part of this repository) to be run separately as a local Uvicorn server.
Verified SafeView Analysis
The application uses `ast.literal_eval` on data retrieved from an external Yahoo Finance MCP server, which could pose a risk if the external server's data is malicious or malformed. However, `ast.literal_eval` is safer than `eval` as it only parses Python literals. `subprocess.Popen` is used to start local server components (like the A2A agent) and is a standard practice for process management. No direct `eval` of untrusted user input or hardcoded sensitive credentials were found. The primary Yahoo Finance MCP server is an external dependency, not part of this codebase. Local network services run on `localhost`.
Updated: 2025-12-01GitHub→
0
0
Low Cost
bhandary icon
Sec8

Provides a Model Context Protocol (MCP) server that enables AI agents to access and manipulate ServiceNow data through a secure API.

Setup Requirements

  • ⚠️Requires a ServiceNow instance URL (SERVICENOW_INSTANCE_URL environment variable or `--url` argument)
  • ⚠️Requires ServiceNow Basic Authentication credentials (SERVICENOW_USERNAME, SERVICENOW_PASSWORD environment variables or `--username`, `--password` arguments)
  • ⚠️Requires the 'mcp' Python library to be installed
Verified SafeView Analysis
No `eval` or obfuscation found. Credentials are primarily retrieved from environment variables or command-line arguments, which is a good practice. The server binds to `0.0.0.0` by default, making it accessible externally if not properly firewalled. Currently, only Basic Authentication is directly supported via the `main` function's command-line parsing, although classes for Token and OAuth authentication exist.
Updated: 2026-01-18GitHub→
0
0
Low Cost
rajsodhi-2 icon

jira-mcp-server

by rajsodhi-2

Sec9

Integrate Claude AI with JIRA for project management, issue tracking, and release planning.

Setup Requirements

  • ⚠️Requires Python 3.8+.
  • ⚠️Requires manual generation of a JIRA API Bearer token with specific 'Read issues', 'Browse projects', and 'View comments' permissions.
  • ⚠️Requires `JIRA_SERVER_URL` and `JIRA_API_TOKEN` environment variables to be set.
Verified SafeView Analysis
The server follows good security practices by explicitly requiring JIRA API tokens and server URLs to be set as environment variables, preventing hardcoding. All communication with the JIRA API is performed over HTTPS. The README advises granting minimal required permissions (Read issues, Browse projects, View comments). No 'eval' or malicious patterns were found. Basic input validation is present for JQL queries and issue keys. There are no explicit rate-limiting or advanced input sanitization features mentioned for all potential JIRA field inputs beyond what JIRA's API provides.
Updated: 2025-11-29GitHub→
0
0
Medium Cost
adamkwhite icon
Sec9

Enables an AI assistant (Claude) to manage Google Workspace services including Calendar, Gmail, and Google Docs through an MCP interface.

Setup Requirements

  • ⚠️Requires manual setup of a Google Cloud Project with specific APIs (Calendar, Gmail, Docs, Drive) enabled.
  • ⚠️Requires downloading OAuth2 credentials (Desktop app type) from Google Cloud Console as 'credentials.json'.
  • ⚠️Requires Python 3.11+ and is designed for optimal compatibility in Ubuntu WSL, which may add environment friction for Windows users.
Verified SafeView Analysis
The server employs robust input validation for metadata fields, including HTML escaping to prevent XSS and URL whitelisting (specifically enforcing `https://*.claude.ai`) to mitigate phishing and data exfiltration risks. A key security design principle is the intentional exclusion of destructive operations (like deletion or permanent removal of events/documents/emails) to prevent accidental data loss through AI interaction. Authentication tokens are stored locally using `pickle.dumps`/`loads`, with a clear justification for `pickle.loads` as it's deserializing trusted, internal data. No direct use of `eval` or other highly dangerous patterns was observed. Overall, it follows a 'safe-by-design' approach.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
klietus icon

SignalZero-MCP

by klietus

Sec8

Serves as an MCP server to manage and access a SignalZero shared symbolic store, wrapping an AWS API Gateway.

Setup Requirements

  • ⚠️Requires an AWS SignalZero API Key (SYMBOL_STORE_API_KEY) for authenticated access to the upstream API.
  • ⚠️Requires a Python 3.x environment with dependencies installed via `pip install -r requirements.txt`.
Verified SafeView Analysis
The server uses API keys for authentication with the upstream SignalZero Symbol Store API, which are loaded from environment variables (SYMBOL_STORE_API_KEY). This is a standard security practice, but proper management of this key is critical. The AWS Lambda function handles requests, with IAM roles providing necessary, but scoped, access to DynamoDB. CORS is broadly enabled for the upstream API. There are no obvious code injection points or uses of 'eval' found in the provided code.
Updated: 2025-11-27GitHub→
0
0
Medium Cost
drewsungg icon

mcpkmn-showdown

by drewsungg

Sec8

An MCP server providing comprehensive competitive Pokémon data (stats, moves, abilities, type matchups, and more) to AI assistants for battle reasoning and agent development.

Setup Requirements

  • ⚠️Requires Python 3.10+.
  • ⚠️Requires specific configuration in Claude Desktop's `mcpServers` settings.
  • ⚠️Pokémon data is cached locally and needs to be manually refreshed using `python -m mcpkmn_showdown.data_fetcher` for the latest updates from Pokémon Showdown.
Verified SafeView Analysis
The server uses `urllib.request.urlopen` to fetch data from `https://play.pokemonshowdown.com/data`. The URL is hardcoded, mitigating risks associated with dynamic URLs. Parsing of minified JavaScript data involves complex regular expressions and `json.loads`. While parsing complex, potentially untrusted text is a theoretical risk, the source data is from a trusted game server, reducing the practical risk of injection or malicious data. No direct `eval` or `exec` calls are present on user-controlled input. No hardcoded secrets are identified.
Updated: 2025-12-13GitHub→
0
0
Medium Cost
HandyS11 icon

ProjGraph

by HandyS11

Sec9

Analyzes .NET solution architectures and generates Entity Relationship Diagrams (ERDs) for AI assistants and developers.

Setup Requirements

  • ⚠️Requires .NET 10.0 SDK or later.
  • ⚠️Requires manual configuration of an MCP-compliant client (e.g., GitHub Copilot, Claude) to integrate the server.
  • ⚠️Tool parameters like 'path' require absolute file system paths.
Verified SafeView Analysis
The server primarily performs static analysis of .NET project and source code files using trusted libraries like Buildalyzer (MSBuild wrapper) and Roslyn. Communication with AI assistants occurs via stdio (Standard Input/Output), limiting direct network exposure. It processes user-provided file paths, which always carries a theoretical risk of malformed inputs, but the underlying parsing tools are robust. No 'eval' or obfuscation is present, and no hardcoded secrets are visible.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
TheZacillac icon

seer

by TheZacillac

Sec4

A multi-interface domain name utility for querying domain registration information, DNS records, and network data, including an MCP server for AI assistant integration.

Setup Requirements

  • ⚠️Requires Rust 1.70+ and Python 3.9+ for full installation.
  • ⚠️Requires `maturin` for building Python bindings.
  • ⚠️The core library used has a documented high-severity SSRF vulnerability.
Review RequiredView Analysis
A high severity Server-Side Request Forgery (SSRF) vulnerability exists in the underlying Rust core library (`seer-core/src/status/client.rs`) that this MCP server depends on. This could allow internal network scanning or access to sensitive cloud metadata if the `status` functionality were exposed or utilized. The `SECURITY_AUDIT.md` explicitly flags this. Additionally, there are several `unwrap()` calls that could lead to panics, and domain validation is inconsistent across some modules. The REST API (part of the same project) lacks HTTP-layer rate limiting.
Updated: 2026-01-16GitHub→
0
0
Medium Cost
patrickoleary icon

vtkapi-mcp

by patrickoleary

Sec9

Provides post-generation validation of VTK Python code to catch API hallucinations via a Model Context Protocol (MCP) server.

Setup Requirements

  • ⚠️Requires 'data/vtk-python-docs.jsonl' (a ~64MB file) to be manually downloaded and placed in the 'data/' directory, or its path explicitly provided via the '--api-docs' argument.
  • ⚠️Python 3.10 or newer is required.
Verified SafeView Analysis
The server operates via standard I/O (stdio), limiting direct network exposure. File access is restricted to a user-specified API documentation file, reducing arbitrary file system risks. No direct 'eval' or 'exec' calls of user-provided input are present. Input schemas for tools are defined, providing a layer of validation, although comprehensive input sanitization for all possible tool inputs is not explicitly detailed in the provided snippets.
Updated: 2025-12-22GitHub→
PreviousPage 664 of 713Next