Data360 MCP Server

A Model Context Protocol (MCP) server that gives LLM agents direct access to the World Bank's Data360 Platform. Agents can search, validate, and retrieve development indicators—covering topics from GDP and poverty to gender equality and climate—without hallucinating data values.

Audience: Developers building AI agents and chatbots that need reliable, structured access to World Bank development data.

Key Features

• Smart indicator discovery — search across hundreds of indicators with enriched metadata and country coverage checks

• Rich metadata retrieval — fetch methodology, definitions, limitations, and statistical concepts on demand

• Reliable time-series data — query historical data points with filters for country, time period, sex, age, and urbanization

• LLM-optimized resources — built-in system prompts, codelists, and chain-of-thought guidance for chatbot integration

• Agent-friendly design — significant "glue" logic makes the raw Data360 API composable and safe for LLM tool use

Related MCP server: World Bank Data360 MCP Server

Getting Started

Prerequisites

• Python 3.11+

• uv (recommended) — or pip

Installation

With uv (recommended):

git clone https://github.com/worldbank/data360-mcp.git
cd data360-mcp
uv sync
# LangChain / LangGraph client + examples + repo-root shim (data360_mcp_service.py):
uv sync --extra agent --group dev

With pip:

git clone https://github.com/worldbank/data360-mcp.git
cd data360-mcp
pip install -e .

Configuration

Copy the example environment file and adjust as needed:

cp .env.example .env

Run the Server

uv run poe serve
# Server starts at http://localhost:8000/mcp

Or with custom port/transport:

uv run poe serve --port 8021 --transport sse
# SSE endpoint: http://localhost:8021/sse

Connect Your Agent

Try the Demo

uv run scripts/llm_mcp_demo.py
# DEBUG mode:
DEBUG=true uv run scripts/llm_mcp_demo.py

Minimal external agent (one-shot): examples/agents/langchain-minimal/README.md — copy-paste run_once.py that loads data360://system-prompt and tools, then calls the model.

Multi-agent / LangGraph: examples/agents/langchain-graph/README.md — register Data360 as a node (create_data360_langgraph_node or gated create_data360_gated_langgraph_node) alongside supervisors and other specialists. Client library (publishable on PyPI): packages/data360-mcp-agent/ (pip install data360-mcp-agent). The repo-root data360_mcp_service.py shim re-exports data360_mcp_agent for older import paths.

MCP Tools

Recommended Agent Workflow

1. Search    → data360_search_indicators(query, required_country="Kenya")
               Returns: covers_country, latest_data, dimensions per indicator

2. Get Data  → data360_get_data(database_id, indicator_id, filters)
               Use REF_AREA code from search; add time period filters

MCP Resources

For chatbot integration, copy data360://system-prompt into your system prompt. It includes chain-of-thought reasoning templates and filter guidance.

Documentation

Project site: worldbank.github.io/data360-mcp — landing page with features, tools, and connection details.

A markdown overview lives in docs/overview.md. The site is deployed with GitHub Actions on pushes to main or dev. In the repository Settings → Pages, set Build and deployment source to GitHub Actions (first-time setup).

Preview locally: from the repository root, run python -m http.server --directory docs and open http://127.0.0.1:8000/.

For developer setup, testing, and contribution instructions, see DEVELOPMENT.md.

Contact

AI for Data - Data for AI Team (ai4data@worldbank.org) Development Data Group / Office of the World Bank Group Chief Statistician World Bank Group

License

This project is licensed under the MIT License together with the World Bank IGO Rider. The Rider is purely procedural: it reserves all privileges and immunities enjoyed by the World Bank, without adding restrictions to the MIT permissions. Please review both files before using, distributing or contributing.

See LICENSE and WB-IGO-RIDER.md for the full license texts.