World Bank Data360 MCP Server

Instructions

    <purpose>
        Search World Bank Data360 for datasets. This is STEP 1 of 3 in the data retrieval workflow.
        Find indicator IDs and database IDs needed for subsequent data operations.
    </purpose>

    <workflow>
        <step number="1">search_datasets (this tool) - Find indicator ID and database ID</step>
        <step number="2">get_temporal_coverage - Check available years BEFORE retrieving data</step>
        <step number="3">retrieve_data - Fetch actual data with proper year and limit parameters</step>
    </workflow>

    <optimization_tips>
        <tip>Remove punctuation: "GDP, total" becomes "GDP total"</tip>
        <tip>Expand abbreviations: "GDP" becomes "Gross Domestic Product"</tip>
        <tip>Add "total" for aggregates: "population" becomes "population total"</tip>
        <tip>Use lowercase for consistency</tip>
        <tip>Remove filler words: "data", "statistics"</tip>
    </optimization_tips>

    <common_databases>
        <database id="WB_WDI">World Development Indicators (most comprehensive)</database>
        <database id="WB_HNP">Health, Nutrition and Population</database>
        <database id="WB_GDF">Global Development Finance</database>
    </common_databases>

    <examples>
        <example original="GDP">gross domestic product total</example>
        <example original="population data">population total</example>
        <example original="">poverty headcount ratio</example>
    </examples>

    <returns>
        List of datasets with indicator IDs, names, database IDs, and search scores.
    </returns>

    <next_step>
        Call get_temporal_coverage with the indicator and database from results.
    </next_step>

Implementation Reference

• src/world_bank_mcp/server.py:365-407 (handler)

  Handler function for the 'search_datasets_tool' tool. Registered with @server.tool() decorator. Includes detailed docstring serving as tool description/schema and calls the core search_datasets helper function.

  @server.tool()
  def search_datasets_tool(search_query: str, top: int = 20) -> dict[str, Any]:
      """[STEP 1/3] Search World Bank Data360 for datasets.
      
      <purpose>
          Search World Bank Data360 for datasets. This is STEP 1 of 3 in the data retrieval workflow.
          Find indicator IDs and database IDs needed for subsequent data operations.
      </purpose>
  
      <workflow>
          <step number="1">search_datasets (this tool) - Find indicator ID and database ID</step>
          <step number="2">get_temporal_coverage - Check available years BEFORE retrieving data</step>
          <step number="3">retrieve_data - Fetch actual data with proper year and limit parameters</step>
      </workflow>
  
      <optimization_tips>
          <tip>Remove punctuation: "GDP, total" becomes "GDP total"</tip>
          <tip>Expand abbreviations: "GDP" becomes "Gross Domestic Product"</tip>
          <tip>Add "total" for aggregates: "population" becomes "population total"</tip>
          <tip>Use lowercase for consistency</tip>
          <tip>Remove filler words: "data", "statistics"</tip>
      </optimization_tips>
  
      <common_databases>
          <database id="WB_WDI">World Development Indicators (most comprehensive)</database>
          <database id="WB_HNP">Health, Nutrition and Population</database>
          <database id="WB_GDF">Global Development Finance</database>
      </common_databases>
  
      <examples>
          <example original="GDP">gross domestic product total</example>
          <example original="population data">population total</example>
          <example original="">poverty headcount ratio</example>
      </examples>
  
      <returns>
          List of datasets with indicator IDs, names, database IDs, and search scores.
      </returns>
  
      <next_step>
          Call get_temporal_coverage with the indicator and database from results.
      </next_step>"""
      return search_datasets(search_query, top)

• src/world_bank_mcp/server.py:43-80 (helper)

  Core helper function implementing the dataset search logic by making a POST request to the World Bank Data360 search API endpoint, processing the response, and formatting results with indicator IDs, names, databases, and search scores.

  def search_datasets(search_query: str, top: int = 20) -> dict[str, Any]:
      """Search World Bank Data360 API"""
      payload = {
          "count": True,
          "select": "series_description/idno, series_description/name, series_description/database_id",
          "search": search_query,
          "top": top,
      }
  
      try:
          response = requests.post(
              SEARCH_ENDPOINT,
              json=payload,
              headers={"Content-Type": "application/json", "Accept": "application/json"},
              timeout=30,
          )
          response.raise_for_status()
          data = response.json()
          
          # Format results nicely
          results = []
          for item in data.get("value", []):
              series = item.get("series_description", {})
              results.append({
                  "indicator": series.get("idno"),
                  "name": series.get("name"),
                  "database": series.get("database_id"),
                  "search_score": round(item.get("@search.score", 0), 2)
              })
          
          return {
              "success": True,
              "total_count": data.get("@odata.count", 0),
              "results": results
          }
          
      except Exception as e:
          return {"success": False, "error": str(e)}