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
dataML007 icon

simple_local_rag

by dataML007

Sec7

A multi-modal Retrieval-Augmented Generation (RAG) system for querying PDF documents with conversation memory via a Streamlit UI, FastAPI backend, and MCP server integration.

Setup Requirements

  • ⚠️Requires an OpenAI API Key (paid service) for embeddings and LLM.
  • ⚠️Requires Python 3.9+ and the UV package manager.
  • ⚠️Requires running three separate services (backend, frontend, MCP server) concurrently.
Verified SafeView Analysis
The system uses `os.getenv` for API keys and recommends storing them in a `.env` file excluded from version control, which is good practice. File uploads are handled with temporary files and explicit `.pdf` extension checks, reducing direct path traversal risks. The `VectorStore` uses `pickle.dump` and `pickle.load` for chunk metadata. While typically used for internal data, if an attacker could tamper with the `chunks.pkl` file, this could lead to a deserialization vulnerability. For a 'local' RAG system, this risk is mitigated by assuming trusted local file access. The FastAPI backend uses `allow_origins=["*"]` for CORS, which is noted as acceptable for local development but a security risk for production deployments.
Updated: 2025-12-06GitHub→
0
0
Low Cost
vishnusrinivas00 icon
Sec8

Analyzes Cloudflare Worker logs stored in DuckDB, providing error summaries and performance metrics via a FastAPI server with MCP compatibility.

Setup Requirements

  • ⚠️Requires Cloudflare API Token, Account ID, and Worker Name to be configured as environment variables for data ingestion.
  • ⚠️Requires a local DuckDB file (cloudflare_logs.duckdb) to be pre-populated with data; the `miletech/fetch_logs.py` script is used for this.
  • ⚠️Python dependencies listed in `requirements.txt` must be installed.
Verified SafeView Analysis
Sensitive credentials (Cloudflare API Token, Account ID, Worker Name) are correctly handled as environment variables via `os.getenv` and `dotenv`. SQL queries in analytics endpoints (`query_error_summary_range`, `query_performance_range`, `query_summary_range`) use parameterized queries, which prevents SQL injection. The `/db/inspect` endpoint uses f-strings for `DESCRIBE` and `SELECT` statements; while the table names are internally derived from `SHOW TABLES`, which limits external injection risk, robust input validation or parameterization would be ideal if table names could ever be influenced by untrusted input. No 'eval' or malicious obfuscation patterns were found. Standard web server security practices should be applied for deployment.
Updated: 2025-12-13GitHub→
0
0
Low Cost
Dipesh600 icon

esewa-mcp-server

by Dipesh600

Sec4

This server acts as a Model Context Protocol (MCP) wrapper for eSewa payment gateway, enabling creation of payment sessions, transaction verification, refunds, and status checks.

Setup Requirements

  • ⚠️Requires secure configuration of `ESEWA_MERCHANT_CODE`, `ESEWA_SECRET_KEY`, and `ESEWA_TOKEN` environment variables, as the provided defaults are test credentials and hardcoded values are a security risk.
  • ⚠️Requires a valid eSewa merchant account for actual payment processing beyond testing.
  • ⚠️The deployed server URL specified in `mcp.json` (`https://YOUR-SERVER-URL.com`) implies a requirement for HTTPS to be configured on the server.
Review RequiredView Analysis
The `smithery.yaml` file hardcodes sensitive environment variables like `ESEWA_SECRET_KEY` and `ESEWA_TOKEN`. Additionally, `services/esewa.js` provides hardcoded fallback values for `ESEWA_MERCHANT_CODE` and `ESEWA_SECRET_KEY` if environment variables are not set. This is a critical security vulnerability as it exposes credentials that should be securely managed and never committed to version control, especially for production deployments. While the code uses `crypto` for HMAC-SHA256, the static keys undermine this security measure.
Updated: 2025-11-28GitHub→
0
0
Medium Cost
vpro1032 icon
Sec2

Provides a standardized Model Context Protocol (MCP) interface for AI assistants to interact with and manage PostgreSQL databases.

Setup Requirements

  • ⚠️Requires Docker and Docker Compose to run the server and its PostgreSQL dependency.
  • ⚠️Database connection details and other configurations must be set via environment variables in a `.env` file.
  • ⚠️Requires an MCP client (e.g., Claude Desktop, VS Code with MCP extension) for interaction.
Review RequiredView Analysis
The server has critical SQL injection vulnerabilities. Table names, schema names, column names, and SQL fragments (like WHERE clauses) are often directly interpolated into SQL queries without proper escaping or identifier quoting. For example, `export_table`, `vacuum_table`, `analyze_table`, `insert_data`, `update_data`, and `delete_data` tools are susceptible. An attacker or a malicious/uncontrolled AI agent could craft inputs to these tools (e.g., in `table`, `schema`, `data` keys, `where`, or `returning` arguments) to execute arbitrary SQL commands like `DROP TABLE` or `DELETE` sensitive data. While `pg` parameterization is used for *values*, it does not protect against injection of SQL identifiers or clauses themselves. No direct use of `eval` or obfuscation was found, and credentials are handled via environment variables, but the SQL injection risk is severe.
Updated: 2025-12-07GitHub→
0
0
Low Cost
takanoriyanagitani icon

rs-cel-mcp

by takanoriyanagitani

Sec8

Provides a Model Context Protocol (MCP) server that enables AI clients to evaluate Common Expression Language (CEL) expressions.

Setup Requirements

  • ⚠️Requires Rust toolchain for building and running.
  • ⚠️Requires an MCP-compatible client (e.g., Gemini CLI, ollmcp) to interact with its tools.
Verified SafeView Analysis
The core functionality involves evaluating user-provided CEL expressions. While CEL is designed for safe, sandboxed evaluation, any evaluator inherently carries a risk if the underlying CEL library has vulnerabilities. The server is designed for local or trusted network usage (e.g., binding to 127.0.0.1 as shown in examples), which mitigates exposure risks. There is no explicit authentication/authorization for the MCP endpoint, which is acceptable for local tools but would be a critical vulnerability if exposed publicly.
Updated: 2025-11-30GitHub→
0
0
Medium Cost
kritsanan1 icon
Sec1

A server component likely designed to manage and process data for a social media application, possibly handling user posts, interactions, and content delivery.

Review RequiredView Analysis
Source code was not provided for analysis. Without the actual source code, it's impossible to perform a security audit for 'eval', hardcoded secrets, network risks, or malicious patterns. Running this server without reviewing its code is extremely risky and strongly discouraged.
Updated: 2025-12-14GitHub→
0
0
Medium Cost
LackOfMorals icon

unofficialMcp

by LackOfMorals

Sec7

The Neo4j MCP server enables LLM clients (e.g., VSCode Copilot Chat, Claude Desktop) to interact with a Neo4j graph database by introspecting its schema and executing Cypher queries.

Setup Requirements

  • ⚠️Requires a running Neo4j database instance with the APOC plugin installed.
  • ⚠️Requires Go 1.25+ to build and run from source.
  • ⚠️Docker is required if running integration tests with the default `USE_CONTAINER=true` configuration.
Verified SafeView Analysis
The server includes a 'write-cypher' tool that allows the execution of arbitrary Cypher queries with write access, which inherently carries the risk of LLM-generated queries causing harm. This is explicitly noted in the documentation, and a 'NEO4J_READ_ONLY=true' environment variable is provided to disable write tools as a mitigation. No other critical vulnerabilities like 'eval' or hardcoded production secrets were identified; default credentials are for local development and are overrideable.
Updated: 2025-12-07GitHub→
0
0
High Cost

crawl4ai-rag-mcp

by AI-enthusiasts

Sec7

Provides a Multi-Crawl Platform for AI-powered Retrieval Augmented Generation (RAG) systems, leveraging agentic search and data analysis.

Setup Requirements

  • ⚠️Docker and Docker Compose are required to run the services and manage their orchestration.
  • ⚠️Requires API keys for external Large Language Models (LLMs) like Claude, which are typically paid services.
  • ⚠️Integration with Searxng implies managing an additional service for search capabilities, potentially requiring specific configuration and maintenance.
Review RequiredView Analysis
The project uses Docker and integrates Searxng, potentially exposing network services. It involves external API calls (e.g., Claude) which require secure credential management. Production deployments will necessitate careful network configuration and secret handling. The presence of `secrets.baseline` indicates some attention to secret management practices.
Updated: 2025-11-18GitHub→
0
0
Low Cost
monolithic-lab icon

modal-mcp-server

by monolithic-lab

Sec8

Provides an API for managing the lifecycle (creation, status, stopping, deletion) of Modal applications.

Setup Requirements

  • ⚠️Requires `modal`, `fastmcp`, and `mcp` Python packages.
  • ⚠️Requires Modal authentication (e.g., `MODAL_TOKEN_ID`, `MODAL_TOKEN_SECRET` environment variables or `modal login`).
  • ⚠️Utilizes Modal's experimental API for app listing, which might require a specific Modal SDK version or be subject to change.
Verified SafeView Analysis
The server exposes an API to manage Modal applications, relying on the security of the Modal SDK and the permissions configured for the underlying Modal account. It generates Python code as templates, but does not execute arbitrary user-provided code directly. Destructive operations (stop, delete) are noted as placeholder implementations within the server, meaning the server itself doesn't directly perform these actions, but rather suggests they are handled by external Modal deployment processes.
Updated: 2025-11-30GitHub→
0
0
Low Cost
TL-Developer icon

mcp-servers

by TL-Developer

Sec8

Defines an MCP server for launching npmlens, a tool likely used for analyzing and managing npm packages.

Setup Requirements

  • ⚠️Requires Node.js and npm/npx to be installed on the system.
Verified SafeView Analysis
The provided source code is a configuration file defining a command to execute an external package via 'npx'. While the configuration itself is benign, running 'npx -y npmlens-mcp@latest' implies trust in the 'npmlens-mcp' package, which could introduce risks if the package itself is compromised or malicious. The '-y' flag bypasses user confirmation.
Updated: 2025-11-19GitHub→
0
0
Low Cost
aelvion icon
Sec8

Orchestrates build, deploy, and infrastructure workflows for applications.

Setup Requirements

  • ⚠️Docker required for local dependency management (Redis, Celery worker, API service).
  • ⚠️Python 3.11 or higher is required.
  • ⚠️Celery and Redis are core dependencies for asynchronous task processing.
Verified SafeView Analysis
The project uses Pydantic for input validation, which helps prevent common injection vulnerabilities. Critical operations like building images, running Terraform, or triggering Jenkins are currently stubbed, which inherently limits immediate execution risks from untrusted input. The `docs/secrets.md` file outlines a strong security posture for credential handling (e.g., IAM Roles Anywhere, OIDC, Kubernetes secrets, avoiding hardcoded secrets). Potential risks would arise if the stubbed tasks were implemented to directly execute shell commands without rigorous input sanitization, or if the `generate-dockerfile` output were directly built by an insecure system, but the current code doesn't exhibit these direct vulnerabilities. Redis is exposed locally via Docker Compose, which is typical for development but would require securing in a production environment.
Updated: 2025-12-01GitHub→
0
0
Medium Cost
icy-r icon
Sec9

An MCP server enabling AI assistants to manage Bitbucket Cloud repositories, pull requests, pipelines, and other related development tasks.

Setup Requirements

  • ⚠️Requires Bitbucket authentication setup (API Token, OAuth, or Basic) which involves setting specific environment variables like BITBUCKET_API_TOKEN and BITBUCKET_USER_EMAIL.
  • ⚠️Node.js version 20.0.0 or higher is required.
  • ⚠️MCP client (e.g., Cursor IDE, Claude Desktop) configuration is necessary to connect to and utilize the server.
Verified SafeView Analysis
The server primarily relies on environment variables for sensitive configuration (API tokens, usernames, passwords, OAuth credentials), which is a good practice. There are no indications of hardcoded secrets or use of 'eval' or other directly exploitable dynamic code execution patterns. Network requests are handled with retry logic and error parsing. Proper configuration of Bitbucket API tokens with least-privilege permissions is crucial for overall security when deploying this server.
Updated: 2025-12-22GitHub→
PreviousPage 524 of 760Next