Gemini MCP Server for Claude Desktop

Instructions

Implementation Reference

• src/tools/image-generation.js:48-116 (handler)

  The main handler function that performs prompt validation and enhancement, calls the Gemini service to generate the image, saves the base64 image data to a PNG file, learns from the interaction, and returns a formatted text response with the file path.
  async execute(args) { const prompt = validateNonEmptyString(args.prompt, 'prompt'); const context = args.context ? validateString(args.context, 'context') : null; log(`Generating image: "${prompt}" with context: ${context || 'general'}`, this.name); try { let enhancedPrompt = prompt; if (this.intelligenceSystem.initialized) { try { enhancedPrompt = await this.intelligenceSystem.enhancePrompt(prompt, context, this.name); log('Applied Tool Intelligence enhancement', this.name); } catch (err) { log(`Tool Intelligence enhancement failed: ${err.message}`, this.name); } } const formattedPrompt = `Create a detailed and high-quality image of: ${enhancedPrompt}`; const imageData = await this.geminiService.generateImage('IMAGE_GENERATION', formattedPrompt); if (imageData) { log('Successfully extracted image data', this.name); ensureDirectoryExists(config.OUTPUT_DIR, this.name); const timestamp = Date.now(); const hash = crypto.createHash('md5').update(prompt).digest('hex'); const imageName = `gemini-${hash}-${timestamp}.png`; const imagePath = path.join(config.OUTPUT_DIR, imageName); fs.writeFileSync(imagePath, Buffer.from(imageData, 'base64')); log(`Image saved to: ${imagePath}`, this.name); if (this.intelligenceSystem.initialized) { try { await this.intelligenceSystem.learnFromInteraction(prompt, enhancedPrompt, `Image generated successfully: ${imagePath}`, context, this.name); log('Tool Intelligence learned from interaction', this.name); } catch (err) { log(`Tool Intelligence learning failed: ${err.message}`, this.name); } } let finalResponse = `✓ Image successfully generated from prompt: "${prompt}"\n\nYou can find the image at: ${imagePath}`; // eslint-disable-line max-len if (context && this.intelligenceSystem.initialized) { finalResponse += `\n\n---\n_Enhancement applied based on context: ${context}_`; } return { content: [ { type: 'text', text: finalResponse, }, ], }; } log('No image data found in response', this.name); return { content: [ { type: 'text', text: `Could not generate image for: "${prompt}". No image data was returned by Gemini API.`, }, ], }; } catch (error) { log(`Error generating image: ${error.message}`, this.name); throw new Error(`Error generating image: ${error.message}`); } }
• src/tools/image-generation.js:22-35 (schema)

  Input schema defining the tool parameters: required 'prompt' string for image description, optional 'context' string for enhancement style.
  { type: 'object', properties: { prompt: { type: 'string', description: 'Text description of the desired image', }, context: { type: 'string', description: 'Optional context for intelligent enhancement (e.g., "artistic", "photorealistic", "technical")', }, }, required: ['prompt'], },
• src/tools/index.js:78-78 (registration)

  Registers a new instance of ImageGenerationTool with the shared intelligenceSystem and geminiService instances in the central tool registry.
  registerTool(new ImageGenerationTool(intelligenceSystem, geminiService));
• src/gemini/gemini-service.js:58-76 (helper)

  Supporting method that interfaces with the Google Gemini API to generate image data (base64) from a text prompt using the IMAGE_GENERATION model configuration.
  async generateImage(modelType, prompt) { try { const modelConfig = getGeminiModelConfig(modelType); // Pass only the model name to getGenerativeModel const model = this.genAI.getGenerativeModel({ model: modelConfig.model }); const content = formatTextPrompt(prompt); // Image generation also uses text prompt // Pass the generationConfig to the generateContent method const result = await model.generateContent({ contents: [{ parts: content }], generationConfig: modelConfig.generationConfig, }); log(`Image generation response received from Gemini API for model type: ${modelType}`, 'gemini-service'); return extractImageData(result.response?.candidates?.[0]); } catch (error) { log(`Error generating image with Gemini API for model type ${modelType}: ${error.message}`, 'gemini-service'); throw new Error(`Gemini image generation failed: ${error.message}`); } }