Skip to main content

MCP server for the Humanitarian Data Exchange (HDX) API

Project description

Humanitarian Data Exchange (HDX) MCP Server

HDX MCP Server used in Claude Desktop
Example of using HDX MCP server used in Claude Desktop

A Model Context Protocol (MCP) server that provides AI assistants access to the Humanitarian Data Exchange (HDX) API. This provides a wide array of humanitarian data hosted on the Humanitarian Data Exchange.

This server can be installed in AI platforms such as Claude Desktop, as well as run as a remote MCP server for any AI to access.

Available Tools

🔍 Metadata & Discovery (12 tools)

  • metadata_location_get - Get countries/locations in HDX
  • metadata_admin1_get - Get states/provinces
  • metadata_admin2_get - Get districts/counties
  • metadata_data_availability_get - Check data coverage (use first!)
  • metadata_dataset_get - Get dataset information
  • metadata_resource_get - Get resource details
  • metadata_org_get - Get organizations
  • metadata_sector_get - Get humanitarian sectors
  • metadata_currency_get - Get currency info
  • metadata_org_type_get - Get organization types
  • metadata_wfp_commodity_get - Get WFP commodities
  • metadata_wfp_market_get - Get WFP markets

🚨 Affected People (4 tools)

  • affected_people_refugees_get - Get refugee/persons of concern data
  • affected_people_humanitarian_needs_get - Get humanitarian needs data
  • affected_people_idps_get - Get internally displaced persons data
  • affected_people_returnees_get - Get returnee data

🌍 Demographics & Geography (1 tool)

  • geography_infrastructure_baseline_population_get - Get population data

🌦️ Climate (1 tool)

  • climate_rainfall_get - Get rainfall data

🤝 Coordination & Context (4 tools)

  • coordination_context_operational_presence_get - Get operational presence
  • coordination_context_funding_get - Get funding data
  • coordination_context_conflict_events_get - Get conflict events
  • coordination_context_national_risk_get - Get national risk data

🍽️ Food Security & Poverty (3 tools)

  • food_security_nutrition_poverty_food_security_get - Get food security data
  • food_security_nutrition_poverty_food_prices_get - Get food prices
  • food_security_nutrition_poverty_poverty_rate_get - Get poverty rates

🔧 Utilities (1 tool)

  • util_version_get - Get API version

🎯 Custom HDX Tools (3 tools)

  • hdx_server_info - Get server information
  • hdx_get_dataset_info - Get detailed dataset info
  • hdx_search_locations - Search for locations

Caveats

The HDX API offers a very rich source of Humanitarian data, but it is complex. The coverage of data by regions differs per country, as showing in the table here. This is not a technical limitation, but due to the challenging work of humanitarian organizations gathering data. This does however introduce some limitations, for example, conflict events in Afghanistan are only available at the province and district level, not national. Therefore questions about total count at the country level should not be supported as this would require gathering large volumes of disaggregated data and summation to the national level, both prohibitive and error-prone.

The MCP server provided prompts for helping LLMs identify and inform the user, but as with any LLM application, evaluation testing and ongoing monitoring is required.

Quick Start (with Docker MCP Tools)

Get a free HDX API 'Application identifier' code ...

  1. Go to the api page
  2. On the 'encode_app_identifier' endpoint, click 'Try it out'
  3. Enter an application name for your process and your email
  4. Click 'Execute'
  5. Make note of the displayed application identifier code

Then install using Docker MCP Hub and MCP Toolkit (recommended) ...

  1. Install Docker Desktop 4.43 or above and start
  2. Activate 'MCP Tools' under Settings > Beta Features
  3. Under 'MCP Tools' search for 'HDX' and add the server
  4. When prompted input your app identifier from above

You can now use with various clients such as Claude Desktop (see 'Clients' section in Docker). To install using Claude Desktop ...

  1. Install Claude Desktop
  2. In Docker 'MCP Toolkit' select Clients and click 'Connect' next to Claude
  3. Restart Claude desktop
  4. Click the little slider icon 'Search and tools' on chat input box
  5. Toggle on 'MCP_DOCKER'

Try asking 'What's the population of Mali on HDX?'.

Other ways to use the HDX MCP Server

UV to run a Remote Server

  1. Install UV (if not already installed): curl -LsSf https://astral.sh/uv/install.sh | sh
  2. Clone this repo: git clone https://github.com/dividor/hdx-mcp.git
  3. cd hdx-mcp
  4. uv sync --all-extras --dev
  5. cp env.example .env
  6. Then Edit your .env and set HDX_APP_IDENTIFIER to the application identifier you created above
  7. Run the server: uv run hdx-mcp-server --transport http --host 0.0.0.0 --port 8000 --verbose

Docker to Run a Remote Server

# Build the Docker image
docker build -t hdx-mcp-server .

# Run with stdio transport (for MCP clients)
docker run --rm -i --env-file .env hdx-mcp-server

# Run with HTTP transport
docker run --rm -p 8000:8000 --env-file .env hdx-mcp-server --transport http --host 0.0.0.0

Testing the server with Inspector

  1. Install MCP Inspector
  2. Set Transport type (top-left) to be: Streamable HTTP
  3. Enter URL: http://127.0.0.1:8000/mcp
  4. Click Connect
  5. Click on Tools or Prompts, and use the get and submit buttons to try out the server

HDX MCP Server in Claude Desktop

The easiest way to use with Claude Desktop is to use DOcker MCP Tools as indicated above. You can also install as follows ...

  1. Complete the basic setup above - ensure the server works with uv run hdx-mcp-server

  2. Get the absolute path to your project:

    cd /path/to/your/hdx-mcp
    pwd  # Copy this path
    
  3. Configure Claude Desktop:

    Open Claude Desktop settings and add to your MCP servers configuration:

    Option A: Using Docker (Recommended)

    {
      "hdx-humanitarian-data": {
        "command": "docker",
        "args": ["run", "--rm", "-i", "--env-file", "/absolute/path/to/your/hdx-mcp2/.env", "hdx-mcp-server"]
      }
    }
    

    Option B: Using UV

    {
      "hdx-humanitarian-data": {
        "command": "uv",
        "args": ["run", "hdx-mcp-server"],
        "cwd": "/absolute/path/to/your/hdx-mcp"
      }
    }
    

    Replace /absolute/path/to/your/hdx-mcp with the actual path from step 2.

  4. Ensure your .env file is configured with your HDX app identifier:

    HDX_APP_IDENTIFIER=your_hdx_app_identifier_here
    HDX_BASE_URL=https://hapi.humdata.org
    
  5. Restart Claude Desktop to load the new MCP server

Once configured, you should see the HDX server appear in Claude's MCP servers list.

Configuration

Environment Variables

The server is configured via environment variables. Copy env.example to .env and customize:

Variable Required Default Description
HDX_APP_IDENTIFIER Yes - Your HDX application identifier
HDX_BASE_URL No https://hapi.humdata.org/api/v2 HDX API base URL
HDX_OPENAPI_URL No https://hapi.humdata.org/openapi.json OpenAPI spec URL
HDX_TIMEOUT No 30.0 HTTP request timeout (seconds)
HDX_APP_NAME No hdx-mcp-server Application name for HDX
HDX_APP_EMAIL No assistant@example.com Contact email for user of HDX API
MCP_HOST No localhost Host for HTTP transport
MCP_PORT No 8000 Port for HTTP transport

Complete Configuration Example

See env.example for a complete configuration template with detailed comments:

# View the configuration template
cat env.example

# Copy and edit for your environment
cp env.example .env
nano .env

Excluded Endpoints

The server automatically excludes the following endpoints:

  • /api/v2/encode_app_identifier - Internal utility for generating app identifiers

Logging

Enable verbose logging for debugging:

uv run hdx-mcp-server --verbose

This will show detailed information about:

  • Server initialization
  • OpenAPI spec loading
  • Tool registration
  • HTTP requests and responses
  • Error details

Contributing

See CONTRIBUTING.md for detailed development guidelines, testing instructions, and contribution workflow.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Related Resources

Support

For issues related to:

Issues & Feedback

  • Issues & Suggestions: Please raise an issue for any bugs, feature requests, or suggestions
  • Security Issues: For security-related concerns, please raise an issue to be contacted privately

Note: This server requires a valid HDX API app identifier. Please ensure you comply with HDX's terms of service and rate limiting guidelines when using this server.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

iflow_mcp_dividor_hdx_mcp_server-1.0.0.tar.gz (470.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

File details

Details for the file iflow_mcp_dividor_hdx_mcp_server-1.0.0.tar.gz.

File metadata

  • Download URL: iflow_mcp_dividor_hdx_mcp_server-1.0.0.tar.gz
  • Upload date:
  • Size: 470.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.10.0 {"installer":{"name":"uv","version":"0.10.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for iflow_mcp_dividor_hdx_mcp_server-1.0.0.tar.gz
Algorithm Hash digest
SHA256 79f06f04707bdd46b91ce1bc24c4d822c7fec9a8f8801075f5d131320c355107
MD5 be6e6cd934ee1ac738c1af8187285013
BLAKE2b-256 050ce9020ab9b282f15bbd3ed4fa49161ea794d46043049e25fc44574fd6f49c

See more details on using hashes here.

File details

Details for the file iflow_mcp_dividor_hdx_mcp_server-1.0.0-py3-none-any.whl.

File metadata

  • Download URL: iflow_mcp_dividor_hdx_mcp_server-1.0.0-py3-none-any.whl
  • Upload date:
  • Size: 29.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.10.0 {"installer":{"name":"uv","version":"0.10.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for iflow_mcp_dividor_hdx_mcp_server-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 0df1d613e9c1b6f3db1a09ce27773a9552f93117775d4ea743565af53cec74a3
MD5 91bb1f72ffb14419e66640cb00a41810
BLAKE2b-256 de81402ffa259e669259e4804a61337f9aa24cb55ac50c15dfb8dc8177933a77

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page