HF Dataset MCP

Instructions

Implementation Reference

• src/tools/get-dataset-size.ts:43-56 (handler)

  The handler function that executes the "get_dataset_size" tool logic.

  async ({ dataset }) => {
    const data = await fetchDatasetViewer<SizeResponse>("/size", {
      dataset,
    });
  
    return {
      content: [
        {
          type: "text" as const,
          text: JSON.stringify(data, null, 2),
        },
      ],
    };
  }

• src/tools/get-dataset-size.ts:5-34 (schema)

  Interface defining the structure of the dataset size response.

  interface SizeResponse {
    size: {
      dataset: {
        num_bytes_original_files?: number;
        num_bytes_parquet_files?: number;
        num_bytes_memory?: number;
        num_rows?: number;
      };
      configs: Array<{
        dataset: string;
        config: string;
        num_bytes_original_files?: number;
        num_bytes_parquet_files?: number;
        num_bytes_memory?: number;
        num_rows?: number;
      }>;
      splits: Array<{
        dataset: string;
        config: string;
        split: string;
        num_bytes_original_files?: number;
        num_bytes_parquet_files?: number;
        num_bytes_memory?: number;
        num_rows?: number;
      }>;
    };
    pending: unknown[];
    failed: unknown[];
    partial: boolean;
  }

• src/tools/get-dataset-size.ts:36-58 (registration)

  Registration function for the "get_dataset_size" tool.

  export function registerGetDatasetSize(server: McpServer) {
    server.tool(
      "get_dataset_size",
      "Get row counts and byte sizes for all configs and splits",
      {
        dataset: z.string().describe("Dataset ID (e.g., 'stanfordnlp/imdb')"),
      },
      async ({ dataset }) => {
        const data = await fetchDatasetViewer<SizeResponse>("/size", {
          dataset,
        });
  
        return {
          content: [
            {
              type: "text" as const,
              text: JSON.stringify(data, null, 2),
            },
          ],
        };
      }
    );
  }