1. Conduid
  2. Developer Tools
  3. mcp-sanctions
MCP server · Developer Tools

mcp-sanctions

Sanctions/watchlist screening (RF, OFAC, EU, UK, UN) with confidence levels for KYC/AML.

Unclaimed MIT last commit a year ago devtools
45Fair

Scored 8 days ago · breakdown

About mcp-sanctions

mcp-sanctions is an MCP server published by madupay in the Developer Tools category: sanctions/watchlist screening (RF, OFAC, EU, UK, UN) with confidence levels for KYC/AML.

The repository has 3 stars and 2 forks, with the last commit a year ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.

Install

uvx
uvx atomno-mcp-sanctions
pip
pip install atomno-mcp-sanctions

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about mcp-sanctions

Powered by Claude · Grounded in docs

I know everything about mcp-sanctions. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

Static checks derived from the repository 7 days ago. Heuristics on the source, not an audit.

  • ✓README presentThe repository ships documentation.
  • ✓License declaredUsable under a stated license.
  • !Tests presentNo tests found. Behaviour changes may go unnoticed between versions.
  • ✓Dependencies pinnedA lockfile pins dependency versions.
  • ✓No dynamic code executionSource scan found no eval, dynamic Function, or shell execution.
  • !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.

README

Sanctions MCP Server

A powerful sanctions screening API MCP Server, leveraging OFAC API to provide comprehensive risk assessments for individuals and organizations.

🎉 Ready for integration with Claude through the Model Context Protocol 🎉

https://github.com/user-attachments/assets/41ea6318-5b17-4da1-8d6d-8cb60b5342ea

Key Features

  • Sanctions Screening: Screen individuals and organizations against major global sanctions lists including OFAC SDN, UN, OFSI, and more
  • Customizable Risk Threshold: Set minimum risk scores to determine match relevance
  • Comprehensive Entity Support: Screen both individuals and organizations with detailed information
  • Multiple Identification Methods: Support for various identifiers including names, addresses, identification documents, and more
  • Source Selection: Choose which sanctions lists to check against based on your compliance needs

How to Use

You can run the Sanctions MCP Server using Node.js directly.

Requirements

  1. Node.js v20 or higher
  2. npm or yarn for dependency management
  3. git

Installation

1. Clone the repository

git clone https://github.com/yourusername/mcp-ofac
cd mcp-ofac

2. Install dependencies

npm install

3. Configure claude_desktop_config.json

Add the following configuration to your claude_desktop_config.json file. Make sure to:

  • Update the paths to match your local Node.js installation and repository location
  • Add your OFAC API key if required
{
  "mcpServers": {
    "sanctions": {
      "command": "/path/to/your/node",
      "args": ["/path/to/your/mcp-ofac/index.js"],
      "env": {
        "OFAC_API_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Usage Examples

Screening an Individual

You can ask Claude to screen individuals against sanctions lists:

"Can you check if John Smith born on 1980-01-15 from Syria is on any sanctions lists?"

Screening an Organization

You can ask Claude to check organizations:

"Is there a sanctioned entity called Tech Solutions Ltd based in Tehran, Iran?"

Custom Screening Parameters

You can specify custom parameters for more targeted screening:

"Check if Maria Rodriguez with passport number AB123456 from Venezuela is on the OFAC SDN list with a minimum match score of 90."

How the Tool Works

The Sanctions MCP Server processes screening requests with the following parameters:

  • cases: Array of entities to screen (individuals or organizations) with identifying information
  • type: Entity type ("person" or "organization")
  • minScore: Minimum score threshold for matches (0-100, default: 95)
  • sources: Which sanctions lists to check against (SDN, NONSDN, DPL, UN, OFSI, etc.)

The tool returns detailed match information, including sanctioned entity details, match scores, and source information.

Data Format

Input Schema

The server accepts detailed entity information including:

  • Name
  • Date of birth (for individuals)
  • Gender (for individuals)
  • Citizenship/Nationality
  • Address details
  • Contact information
  • Identification documents

Response Format

The server returns a JSON response with:

  • Match details
  • Match scores
  • Source information
  • Entity profiles

Restart Claude Desktop

After configuring your claude_desktop_config.json, restart Claude Desktop to apply the changes and enable the sanctions screening capability.

README fetched from the repository 7 days ago. The original is authoritative.

Questions

About mcp-sanctions

How do I install mcp-sanctions?

Run uvx atomno-mcp-sanctions, then add the server to your MCP client's configuration.

Is mcp-sanctions safe to use with an AI agent?

Its trust score is 45 out of 100 (fair). It passes 4 of the 6 static security checks run so far; the failures are listed above. It has no ConduID identity yet, so agent calls to it are not receipted.

Is mcp-sanctions still maintained?

The last commit was a year ago. That's long enough that you should check whether the maintainer is responding to issues before depending on it.