Nano-Banana-MCP

Instructions

Implementation Reference

• src/index.ts:214-301 (handler)

  The primary handler function for the generate_image tool. It validates configuration, calls the Gemini API to generate an image from the prompt, processes the response (saving the image file and extracting inline data), builds a formatted text response with file paths and instructions, and returns the MCP tool result.

  private async generateImage(request: CallToolRequest): Promise<CallToolResult> {
    if (!this.ensureConfigured()) {
      throw new McpError(ErrorCode.InvalidRequest, "Gemini API token not configured. Use configure_gemini_token first.");
    }
  
    const { prompt } = request.params.arguments as { prompt: string };
    
    try {
      const response = await this.genAI!.models.generateContent({
        model: "gemini-2.5-flash-image-preview",
        contents: prompt,
      });
      
      // Process response to extract image data
      const content: any[] = [];
      const savedFiles: string[] = [];
      let textContent = "";
      
      // Get appropriate save directory based on OS
      const imagesDir = this.getImagesDirectory();
      
      // Create directory
      await fs.mkdir(imagesDir, { recursive: true, mode: 0o755 });
      
      if (response.candidates && response.candidates[0]?.content?.parts) {
        for (const part of response.candidates[0].content.parts) {
          // Process text content
          if (part.text) {
            textContent += part.text;
          }
          
          // Process image data
          if (part.inlineData?.data) {
            const timestamp = new Date().toISOString().replace(/[:.]/g, '-');
            const randomId = Math.random().toString(36).substring(2, 8);
            const fileName = `generated-${timestamp}-${randomId}.png`;
            const filePath = path.join(imagesDir, fileName);
            
            const imageBuffer = Buffer.from(part.inlineData.data, 'base64');
            await fs.writeFile(filePath, imageBuffer);
            savedFiles.push(filePath);
            this.lastImagePath = filePath;
            
            // Add image to MCP response
            content.push({
              type: "image",
              data: part.inlineData.data,
              mimeType: part.inlineData.mimeType || "image/png",
            });
          }
        }
      }
      
      // Build response content
      let statusText = `🎨 Image generated with nano-banana (Gemini 2.5 Flash Image)!\n\nPrompt: "${prompt}"`;
      
      if (textContent) {
        statusText += `\n\nDescription: ${textContent}`;
      }
      
      if (savedFiles.length > 0) {
        statusText += `\n\n📁 Image saved to:\n${savedFiles.map(f => `- ${f}`).join('\n')}`;
        statusText += `\n\n💡 View the image by:`;
        statusText += `\n1. Opening the file at the path above`;
        statusText += `\n2. Clicking on "Called generate_image" in Cursor to expand the MCP call details`;
        statusText += `\n\n🔄 To modify this image, use: continue_editing`;
        statusText += `\n📋 To check current image info, use: get_last_image_info`;
      } else {
        statusText += `\n\nNote: No image was generated. The model may have returned only text.`;
        statusText += `\n\n💡 Tip: Try running the command again - sometimes the first call needs to warm up the model.`;
      }
      
      // Add text content first
      content.unshift({
        type: "text",
        text: statusText,
      });
      
      return { content };
      
    } catch (error) {
      console.error("Error generating image:", error);
      throw new McpError(
        ErrorCode.InternalError,
        `Failed to generate image: ${error instanceof Error ? error.message : String(error)}`
      );
    }
  }

• src/index.ts:71-84 (registration)

  Registration of the generate_image tool in the ListToolsRequestSchema handler, including name, description, and input schema.

  {
    name: "generate_image",
    description: "Generate a NEW image from text prompt. Use this ONLY when creating a completely new image, not when modifying an existing one.",
    inputSchema: {
      type: "object",
      properties: {
        prompt: {
          type: "string",
          description: "Text prompt describing the NEW image to create from scratch",
        },
      },
      required: ["prompt"],
    },
  },

• src/index.ts:74-83 (schema)

  Input schema definition for the generate_image tool, specifying a required 'prompt' string parameter.

  inputSchema: {
    type: "object",
    properties: {
      prompt: {
        type: "string",
        description: "Text prompt describing the NEW image to create from scratch",
      },
    },
    required: ["prompt"],
  },

• src/index.ts:159-160 (handler)

  Dispatcher case in the CallToolRequestSchema handler that routes generate_image calls to the generateImage method.

  case "generate_image":
    return await this.generateImage(request);