OECD

You can access OECD’s vast SDMX data through a Model Context Protocol (MCP) server that connects AI assistants to 5,000+ economic and statistical datasets from 38 OECD economies. Use natural language queries to search, compare, and analyze data on economy, health, education, environment, and more, all via MCP-compatible clients.

How to use

Connect an MCP client to the remote OECD MCP server to start querying datasets and generating insights. You can use HTTP-based MCP access for remote operation or run a local MCP server instance to keep data interactions entirely within your environment.

How to install

Prerequisites you need before installation:

• Node.js (recommended) and npm for building and running local MCP server
• Access to a terminal or command prompt
• Basic familiarity with JSON/TOML configuration

Option A: Run the remote OECD MCP server (no local build required)

There is a remote MCP endpoint you can connect to from your MCP client. Use this URL in your client configuration to access OECD data.

Option B: Run a local instance from source (requires building)

1. Clone the repository and install dependencies

git clone https://github.com/isakskogstad/OECD-MCP.git
cd OECD-MCP
npm install
npm run build

2. Run the local server (if your setup requires starting the built index)

node dist/index.js

3. Configure local clients to point to the local MCP endpoint if you are running the server locally.

Additional configuration and usage notes

Connection options include a remote HTTP MCP URL and a local stdio transport. The remote URL is https://oecd-mcp.onrender.com/mcp. For local use, the MCP server can be started with a standard Node invocation that runs the built index.

{
  "mcpServers": {
    "oecd": {
      "command": "node",
      "args": ["/absolute/path/to/OECD-MCP/dist/index.js"]
    }
  }
}

Security and access notes

If you are using the remote MCP URL, ensure you secure any API keys or tokens required by your client. When running locally, protect your environment and avoid exposing internal paths or server ports publicly.

All data access is subject to OECD SDMX API usage terms and conditions. Review those terms to ensure compliant usage in your applications.

Troubleshooting tips

If you cannot reach the remote MCP endpoint, verify network access and that your client is configured to connect to https://oecd-mcp.onrender.com/mcp. For local runs, confirm that the started process is listening on the expected port and that the path to dist/index.js is correct.

Available tools

search_dataflows

Search for datasets by keyword to discover relevant dataflows.

list_dataflows

List datasets by category to browse available data.

get_categories

List all 17 data categories available in the OECD data collection.

get_popular_datasets

Retrieve commonly used datasets for quick access.

search_indicators

Search for specific indicators across datasets.

get_data_structure

Access metadata describing dataset dimensions and structure.

query_data

Query statistical data from selected datasets with filters.

get_dataflow_url

Generate a link to the OECD Data Explorer for a dataset.

list_categories_detailed

Retrieve detailed information about each data category.