MCP-wolfram-alpha

Instructions

Implementation Reference

• src/mcp_wolfram_alpha/server.py:89-125 (handler)

  Executes the 'query-wolfram-alpha' tool: calls the Wolfram client with the query, processes the response pods and subpods to extract plaintext and images (downloading and base64 encoding images), and returns a list of TextContent or ImageContent.

  if name == "query-wolfram-alpha":
      results: list[types.TextContent | types.ImageContent | types.EmbeddedResource] = []
      query = arguments.get("query")
      if not query:
          raise ValueError("Missing 'query' parameter for Wolfram Alpha tool")
  
      try:
          response = await client.aquery(query)
      except Exception as e:
          raise Exception("Failed to query Wolfram Alpha") from e
      
      try:
          async with httpx.AsyncClient() as http_client:
              for pod in response.pods:
                  for subpod in pod.subpods:
  
                      if subpod.plaintext:  # Handle text content
                          results.append(types.TextContent(
                              type="text",
                              text=subpod.plaintext
                          ))
                          
                      elif subpod.img:  # Handle image content
                          img_url = subpod.img.get("src")
                          if img_url:
                              img_response = await http_client.get(img_url)
                              if img_response.status_code == 200:
                                  img_base64 = base64.b64encode(img_response.content).decode('utf-8')
                                  results.append(types.ImageContent(
                                      type="image",
                                      data=img_base64,
                                      mimeType="image/png"
                                  ))
      except Exception as e:
          raise Exception("Failed to parse response from Wolfram Alpha") from e
  
      return results

• src/mcp_wolfram_alpha/server.py:64-75 (registration)

  Registers the 'query-wolfram-alpha' tool in the list_tools() method, specifying its name, description, and input JSON schema.

      types.Tool(
          name="query-wolfram-alpha",
          description="Use Wolfram Alpha to answer a question. This tool should be used when you need complex math or symbolic intelligence.",
          inputSchema={
              "type": "object",
              "properties": {
                  "query": {"type": "string"}  # Correct property: `query` with type `string`
              },
              "required": ["query"]  # Marking `query` as required
          },
      )
  ]

• src/mcp_wolfram_alpha/server.py:67-73 (schema)

  JSON schema for tool input: requires a single 'query' property of type string.

  inputSchema={
      "type": "object",
      "properties": {
          "query": {"type": "string"}  # Correct property: `query` with type `string`
      },
      "required": ["query"]  # Marking `query` as required
  },

• src/mcp_wolfram_alpha/server.py:6-6 (helper)

  Imports the WolframAlpha client used by the tool handler to perform the actual API query.

  from .wolfram_client import client