Multi-Purpose MCP Server

Instructions

Input Schema

Input Schema (JSON Schema)

Implementation Reference

• src/index.ts:256-289 (handler)

  Core handler function that generates an image from a text prompt using Hugging Face Inference API (FLUX.1-schnell model) and returns it as base64 string.
  const generateImage = async (prompt: string, hfToken: string): Promise<string> => { if (!hfToken) { throw new Error('HF_TOKEN이 설정되지 않았습니다. Hugging Face API 토큰을 설정해주세요.') } const client = new InferenceClient(hfToken) try { const image = await client.textToImage({ provider: "fal-ai", model: "black-forest-labs/FLUX.1-schnell", inputs: prompt, parameters: { num_inference_steps: 5 } }) // Convert to base64 let base64: string if (typeof image === 'string') { // If it's already a base64 string, use it directly base64 = image } else if (image && typeof image === 'object' && 'arrayBuffer' in image) { // If it's a Blob-like object const arrayBuffer = await (image as any).arrayBuffer() base64 = Buffer.from(arrayBuffer).toString('base64') } else { // If it's a Buffer or Uint8Array base64 = Buffer.from(image as any).toString('base64') } return base64 } catch (error) { throw new Error(`이미지 생성 중 오류가 발생했습니다: ${error instanceof Error ? error.message : '알 수 없는 오류'}`) } }
• src/index.ts:39-41 (schema)

  Zod schema defining the input for generate_image tool: a single 'prompt' string parameter.
  const ImageGenerationToolSchema = z.object({ prompt: z.string().describe('이미지 생성을 위한 프롬프트') })
• src/index.ts:394-407 (registration)

  Registration of generate_image tool in the MCP listTools handler, specifying name, description, and input schema.
  { name: 'generate_image', description: '텍스트 프롬프트를 사용하여 이미지를 생성하는 도구', inputSchema: { type: 'object', properties: { prompt: { type: 'string', description: '이미지 생성을 위한 프롬프트' } }, required: ['prompt'] } }
• src/index.ts:579-608 (handler)

  MCP CallToolRequest handler branch for generate_image: validates input with schema, invokes generateImage function, returns base64 image content or error text.
  if (request.params.name === 'generate_image') { try { const { prompt } = ImageGenerationToolSchema.parse(request.params.arguments) const base64Image = await generateImage(prompt, hfToken) return { content: [ { type: 'image', data: base64Image, mimeType: 'image/png', annotations: { audience: ['user'], priority: 0.9 } } ] } } catch (error) { return { content: [ { type: 'text', text: `이미지 생성 오류: ${error instanceof Error ? error.message : '알 수 없는 오류'}` } ], isError: true } }