Langfuse MCP Server

Instructions

Implementation Reference

• src/tools/get-dataset.ts:10-26 (handler)

  The main async handler function that implements the core logic of the 'get_dataset' tool. It fetches the dataset from the Langfuse client using the provided datasetName and returns formatted JSON content or an error response.

  export async function getDataset(
    client: LangfuseAnalyticsClient,
    args: GetDatasetArgs
  ) {
    try {
      const data = await client.getDataset(args.datasetName);
      return {
        content: [{ type: 'text' as const, text: JSON.stringify(data, null, 2) }],
      };
    } catch (error) {
      const errorMessage = error instanceof Error ? error.message : String(error);
      return {
        content: [{ type: 'text' as const, text: `Error: ${errorMessage}` }],
        isError: true,
      };
    }
  }

• src/tools/get-dataset.ts:4-6 (schema)

  Zod schema defining the input parameters for the 'get_dataset' tool, specifically requiring a non-empty datasetName string.

  export const getDatasetSchema = z.object({
    datasetName: z.string().min(1).describe('Name of the dataset to retrieve'),
  });

• src/index.ts:662-675 (registration)

  Tool registration in the allTools array used by listToolsRequestHandler. Defines the tool's name, description, and input schema for MCP protocol advertisement.

  {
    name: 'get_dataset',
    description: 'Get detailed information about a specific dataset by name.',
    inputSchema: {
      type: 'object',
      properties: {
        datasetName: {
          type: 'string',
          description: 'Name of the dataset to retrieve',
        },
      },
      required: ['datasetName'],
    },
  },

• src/index.ts:1103-1106 (handler)

  Dispatch handler in the main CallToolRequestSchema switch statement that parses arguments using the schema and invokes the getDataset handler function.

  case 'get_dataset': {
    const args = getDatasetSchema.parse(request.params.arguments);
    return await getDataset(this.client, args);
  }