run_tests

Instructions

Implementation Reference

• Server~/src/tools/runTestsTool.ts:56-108 (handler)

  Core execution logic for the 'run_tests' tool: parses parameters, sends request to McpUnity to run Unity Test Runner tests, handles response, extracts statistics and results, returns formatted text content with summary and JSON details.

  async function toolHandler(mcpUnity: McpUnity, params: any = {}): Promise<CallToolResult> {
    const {
      testMode = 'EditMode',
      testFilter = '',
      returnOnlyFailures = true,
      returnWithLogs = false
    } = params;
  
    // Create and wait for the test run
    const response = await mcpUnity.sendRequest({
      method: toolName,
      params: { 
        testMode,
        testFilter,
        returnOnlyFailures,
        returnWithLogs
      }
    });
    
    // Process the test results
    if (!response.success) {
      throw new McpUnityError(
        ErrorType.TOOL_EXECUTION,
        response.message || `Failed to run tests: Mode=${testMode}, Filter=${testFilter || 'none'}`
      );
    }
    
    // Extract test results
    const testResults = response.results || [];
    const testCount = response.testCount || 0;
    const passCount = response.passCount || 0;
    const failCount = response.failCount || 0;
    const skipCount = response.skipCount || 0;
    
    return {
      content: [
        {
          type: 'text',
          text: response.message
        },
        {
          type: 'text',
          text: JSON.stringify({
            testCount,
            passCount,
            failCount,
            skipCount,
            results: testResults
          }, null, 2)
        }
      ]
    };
  }

• Server~/src/tools/runTestsTool.ts:11-16 (schema)

  Zod schema defining the input parameters for the 'run_tests' tool.

  const paramsSchema = z.object({
    testMode: z.string().optional().default('EditMode').describe('The test mode to run (EditMode or PlayMode) - defaults to EditMode (optional)'),
    testFilter: z.string().optional().default('').describe('The specific test filter to run (e.g. specific test name or class name, must include namespace) (optional)'),
    returnOnlyFailures: z.boolean().optional().default(true).describe('Whether to show only failed tests in the results (optional)'),
    returnWithLogs: z.boolean().optional().default(false).describe('Whether to return the test logs in the results (optional)')
  });

• Server~/src/tools/runTestsTool.ts:26-46 (registration)

  Registration function for the 'run_tests' tool, which calls server.tool with name, description, schema, and a logging wrapper around the toolHandler.

  export function registerRunTestsTool(server: McpServer, mcpUnity: McpUnity, logger: Logger) {
    logger.info(`Registering tool: ${toolName}`);
    
    // Register this tool with the MCP server
    server.tool(
      toolName,
      toolDescription,
      paramsSchema.shape,
      async (params: any = {}) => {
        try {
          logger.info(`Executing tool: ${toolName}`, params);
          const result = await toolHandler(mcpUnity, params);
          logger.info(`Tool execution successful: ${toolName}`);
          return result;
        } catch (error) {
          logger.error(`Tool execution failed: ${toolName}`, error);
          throw error;
        }
      }
    );
  }

• Server~/src/index.ts:57-57 (registration)

  Application-level call to register the 'run_tests' tool during server initialization.

  registerRunTestsTool(server, mcpUnity, toolLogger);