Gemini MCP

Instructions

Implementation Reference

• src/tools/registry.js:85-104 (handler)

  The handler function that implements the core logic of the 'mcp__gemini__code_analyze' tool. It destructures arguments, validates input, constructs an analysis prompt, calls the AI client for code review, and formats the response.

        async (args) => {
          const { code, language = 'javascript', analysis_type = 'comprehensive' } = args;
          validateString(code, 'code', 10000);
          
          const prompt = `Analyze this ${language} code for ${analysis_type} review:
  
  \`\`\`${language}
  ${code}
  \`\`\`
  
  Provide:
  1. Code quality assessment
  2. Security vulnerabilities
  3. Performance issues
  4. Best practice violations
  5. Specific improvement suggestions`;
  
          const analysis = await aiClient.call(prompt, 'analysis');
          return `📊 **Code Analysis** (${analysis_type})\\n\\n${analysis}`;
        }

• src/tools/registry.js:77-105 (registration)

  The registration of the 'mcp__gemini__code_analyze' tool in the ToolRegistry class, specifying name, description, input parameters, and the handler function.

      this.registerTool(
        'mcp__gemini__code_analyze',
        'Analyze code for quality and issues',
        {
          code: { type: 'string', description: 'Code to analyze', required: true },
          language: { type: 'string', description: 'Programming language' },
          analysis_type: { type: 'string', description: 'Type of analysis', default: 'comprehensive' }
        },
        async (args) => {
          const { code, language = 'javascript', analysis_type = 'comprehensive' } = args;
          validateString(code, 'code', 10000);
          
          const prompt = `Analyze this ${language} code for ${analysis_type} review:
  
  \`\`\`${language}
  ${code}
  \`\`\`
  
  Provide:
  1. Code quality assessment
  2. Security vulnerabilities
  3. Performance issues
  4. Best practice violations
  5. Specific improvement suggestions`;
  
          const analysis = await aiClient.call(prompt, 'analysis');
          return `📊 **Code Analysis** (${analysis_type})\\n\\n${analysis}`;
        }
      );

• src/tools/registry.js:80-83 (schema)

  The input parameters object used to generate the JSON schema for the tool's arguments: code (required string), language (string), analysis_type (string with default).

  {
    code: { type: 'string', description: 'Code to analyze', required: true },
    language: { type: 'string', description: 'Programming language' },
    analysis_type: { type: 'string', description: 'Type of analysis', default: 'comprehensive' }