Meilisearch MCP Server

Instructions

Implementation Reference

• src/tools/vector-tools.ts:25-36 (handler)

  The handler function that executes the tool logic: POSTs to Meilisearch /experimental-features endpoint to enable vectorStore experimental feature and returns the API response or handles errors.

  async () => {
    try {
      const response = await apiClient.post('/experimental-features', {
        vectorStore: true,
      });
      return {
        content: [{ type: "text", text: JSON.stringify(response.data, null, 2) }],
      };
    } catch (error) {
      return createErrorResponse(error);
    }
  }

• src/tools/vector-tools.ts:21-37 (registration)

  Direct registration of the 'enable-vector-search' MCP tool, including name, description, empty input schema, and inline handler.

  server.tool(
    "enable-vector-search",
    "Enable the vector search experimental feature in Meilisearch",
    {},
    async () => {
      try {
        const response = await apiClient.post('/experimental-features', {
          vectorStore: true,
        });
        return {
          content: [{ type: "text", text: JSON.stringify(response.data, null, 2) }],
        };
      } catch (error) {
        return createErrorResponse(error);
      }
    }
  );

• src/index.ts:68-68 (registration)

  Top-level registration call that invokes registerVectorTools to add the 'enable-vector-search' tool (among others) to the main MCP server instance.

  registerVectorTools(server);

• src/index.ts:11-11 (registration)

  Import of the vector-tools module containing the enable-vector-search tool registration.

  import registerVectorTools from './tools/vector-tools.js';