Ultra MCP

Instructions

Implementation Reference

• src/server.ts:295-302 (registration)

  Registration of the 'review-code' tool in the MCP server, specifying title, description, input schema (ReviewCodeSchema), and handler function that delegates to AIToolHandlers.handleReviewCode

  server.registerTool("review-code", {
    title: "Review Code",
    description: "Review code for bugs, security issues, performance, or style problems",
    inputSchema: ReviewCodeSchema.shape,
  }, async (args) => {
    const aiHandlers = await getHandlers();
    return await aiHandlers.handleReviewCode(args);
  });

• src/server.ts:41-46 (schema)

  Zod schema definition for 'review-code' tool input validation: task, optional files, focus (bugs/security/performance/style/all), provider

  const ReviewCodeSchema = z.object({
    task: z.string().describe("What to review (e.g., 'review pull request changes', 'check for security issues')"),
    files: z.array(z.string()).optional().describe("File paths to review (optional)"),
    focus: z.enum(["bugs", "security", "performance", "style", "all"]).default("all").describe("Review focus area"),
    provider: z.enum(["openai", "gemini", "azure", "grok"]).optional().default("gemini").describe("AI provider to use"),
  });

• src/handlers/ai-tools.ts:350-398 (handler)

  Core handler implementation in AIToolHandlers class: selects provider, builds focus-specific system prompt for code review, generates AI response with low temperature, returns structured content and metadata

  async handleReviewCode(params: z.infer<typeof ReviewCodeSchema>) {
    // Use provided provider or get the preferred one (Azure if configured)
    const providerName = params.provider || (await this.providerManager.getPreferredProvider(['openai', 'gemini', 'azure', 'grok']));
    const provider = await this.providerManager.getProvider(providerName);
    
    const focusPrompts = {
      bugs: "Focus on identifying potential bugs, logic errors, and runtime issues",
      security: "Focus on security vulnerabilities, input validation, and secure coding practices",
      performance: "Focus on performance bottlenecks, inefficient algorithms, and optimization opportunities",
      style: "Focus on code style, formatting, naming conventions, and readability",
      all: "Provide comprehensive code review covering bugs, security, performance, and style"
    };
  
    const systemPrompt = `You are an expert code reviewer. Review the provided code thoroughly.
    ${focusPrompts[params.focus]}
    
    Provide detailed feedback on:
    - Issues found and their severity
    - Specific recommendations for improvement
    - Code quality assessment
    - Best practices and standards compliance
    
    Be constructive and specific in your review comments.`;
  
    const prompt = `Review the following: ${params.task}${params.files ? `\n\nFiles to review: ${params.files.join(", ")}` : ""}`;
  
    const response = await provider.generateText({
      prompt,
      systemPrompt,
      temperature: 0.2, // Very low temperature for code review accuracy
      reasoningEffort: (providerName === "openai" || providerName === "azure" || providerName === "grok") ? "high" : undefined,
      useSearchGrounding: false, // No search needed for code review
    });
  
    return {
      content: [
        {
          type: "text",
          text: response.text,
        },
      ],
      metadata: {
        provider: providerName,
        model: response.model,
        focus: params.focus,
        usage: response.usage,
        ...response.metadata,
      },
    };