BugBug MCP Server

Instructions

Implementation Reference

• src/tools/testRuns.ts:168-221 (handler)

  Full tool definition including the handler function that fetches and lists screenshots from a test run using the bugbugClient. This is the core implementation of the tool logic.

  export const getTestRunScreenshotsTool: Tool = {
    name: 'get_test_run_screenshots',
    title: 'Get screenshots from a BugBug test run',
    description: 'Get screenshots from a BugBug test run',
    inputSchema: z.object({
      runId: z.string().describe('Test run UUID'),
    }).shape,
    handler: async ({ runId }) => {
        try {
  
          const response = await bugbugClient.getTestRunScreenshots(runId);
          
          if (response.status !== 200) {
            return {
              content: [
                {
                  type: 'text',
                  text: `Error: ${response.status} ${response.statusText}`,
                },
              ],
            };
          }
  
          const screenshots = response.data;
          
          let screenshotsList = '';
          if (screenshots.stepsRuns && screenshots.stepsRuns.length > 0) {
            screenshotsList = screenshots.stepsRuns.map((step: BugBugStepDetail) => 
              `- **Step ${step.stepId}:** ${step.screenshots?.[0]?.url || 'No screenshot'}`
            ).join('\n');
          } else {
            screenshotsList = 'No screenshots available';
          }
          
          return {
            content: [
              {
                type: 'text',
                text: `**Test Run Screenshots (ID: ${screenshots.id}):**\n\n${screenshotsList}`,
              },
            ],
          };
        } catch (error) {
          return {
            content: [
              {
                type: 'text',
                text: `Error fetching test run screenshots: ${error instanceof Error ? error.message : 'Unknown error'}`,
              },
            ],
          };
        }
      }
  };

• src/tools/testRuns.ts:172-174 (schema)

  Zod input schema for the tool, requiring a runId string.

  inputSchema: z.object({
    runId: z.string().describe('Test run UUID'),
  }).shape,

• src/tools/index.ts:11-33 (registration)

  Registers all tools on the MCP server, including those from testRunsTools (which exports getTestRunScreenshotsTool) by iterating over the tools record and calling server.registerTool.

  export function registerAllTools(server: McpServer): void {
    const tools: Record<string, Tool> = {
      ...configTools,
      ...testsTools,
      ...testRunsTools,
      ...suitesTools,
      ...suiteRunsTools,
      ...profilesTools,
      ...advancedTools,
    };
  
    for (const t in tools) {
      server.registerTool(
        tools[t].name,
        {
          description: tools[t].description,
          inputSchema: tools[t].inputSchema,
          annotations: { title: tools[t].title },
        },
        (args: unknown) => tools[t].handler(args as unknown)
      );
    }
  }

• src/services/bugbugClient.ts:238-240 (helper)

  The bugbugClient method called by the tool handler to make the API request for test run screenshots.

  async getTestRunScreenshots(id: string): Promise<ApiResponse<BugBugScreenshotResponse>> {
    return this.makeRequest(`/testruns/${id}/screenshots/`);
  }