Draw Things MCP Server

Instructions

Implementation Reference

• src/tools/transform-image.ts:55-159 (handler)

  The handler function that validates parameters, checks server status, reads the source image if needed, calls the DrawThingsClient img2img method to transform the image, saves the result, and returns a JSON summary.

  export async function transformImage(
    client: DrawThingsClient,
    params: z.infer<typeof transformImageSchema>
  ): Promise<{ type: "text"; text: string }[]> {
    try {
      // Validate input
      if (!params.image_path && !params.image_base64) {
        return [
          {
            type: "text",
            text: "Error: Either image_path or image_base64 must be provided",
          },
        ];
      }
  
      // Check if server is running first
      const status = await client.checkStatus();
      if (!status.running) {
        return [
          {
            type: "text",
            text: `Error: ${status.message}`,
          },
        ];
      }
  
      // Get base64 image data
      let imageBase64 = params.image_base64;
      if (params.image_path && !imageBase64) {
        try {
          imageBase64 = await client.readImageAsBase64(params.image_path);
        } catch (error) {
          const message = error instanceof Error ? error.message : String(error);
          return [
            {
              type: "text",
              text: `Error reading image file: ${message}`,
            },
          ];
        }
      }
  
      // Transform the image
      const response = await client.img2img({
        prompt: params.prompt,
        init_images: [imageBase64!],
        negative_prompt: params.negative_prompt,
        denoising_strength: params.denoising_strength,
        steps: params.steps,
        cfg_scale: params.cfg_scale,
        seed: params.seed,
      });
  
      if (!response.images || response.images.length === 0) {
        return [
          {
            type: "text",
            text: "Error: No images were generated",
          },
        ];
      }
  
      // Save the image(s)
      const savedPaths: string[] = [];
      for (let i = 0; i < response.images.length; i++) {
        const outputPath =
          params.output_path && response.images.length === 1
            ? params.output_path
            : undefined;
        const path = await client.saveImage(response.images[i], outputPath);
        savedPaths.push(path);
      }
  
      return [
        {
          type: "text",
          text: JSON.stringify(
            {
              success: true,
              message: `Transformed image, generated ${savedPaths.length} result(s)`,
              files: savedPaths,
              prompt: params.prompt,
              source: params.image_path || "(base64 input)",
              parameters: {
                denoising_strength: params.denoising_strength ?? 0.75,
                steps: params.steps || 20,
                cfg_scale: params.cfg_scale || 7.5,
                seed: params.seed ?? -1,
              },
            },
            null,
            2
          ),
        },
      ];
    } catch (error) {
      const message = error instanceof Error ? error.message : String(error);
      return [
        {
          type: "text",
          text: `Error transforming image: ${message}`,
        },
      ];
    }
  }

• src/tools/transform-image.ts:4-50 (schema)

  Zod schema defining the input parameters for the transform_image tool, including prompt, image source (path or base64), optional negative prompt, denoising strength, steps, CFG scale, seed, and output path.

  export const transformImageSchema = z.object({
    prompt: z
      .string()
      .describe("Text description of the desired transformation"),
    image_path: z
      .string()
      .optional()
      .describe("Path to the source image file to transform"),
    image_base64: z
      .string()
      .optional()
      .describe("Base64-encoded source image (alternative to image_path)"),
    negative_prompt: z
      .string()
      .optional()
      .describe("Elements to exclude from the transformed image"),
    denoising_strength: z
      .number()
      .min(0)
      .max(1)
      .optional()
      .describe(
        "Strength of the transformation (0.0-1.0, default: 0.75). Lower values keep more of the original image."
      ),
    steps: z
      .number()
      .int()
      .min(1)
      .max(150)
      .optional()
      .describe("Number of inference steps (default: 20)"),
    cfg_scale: z
      .number()
      .min(1)
      .max(30)
      .optional()
      .describe("Classifier-free guidance scale (default: 7.5)"),
    seed: z
      .number()
      .int()
      .optional()
      .describe("Random seed for reproducibility (-1 for random)"),
    output_path: z
      .string()
      .optional()
      .describe("Custom file path to save the transformed image"),
  });

• src/index.ts:67-75 (registration)

  Registration of the transform_image tool on the MCP server, importing the handler, schema, and description from the tool module and wrapping the handler call.

  server.tool(
    "transform_image",
    transformImageDescription,
    transformImageSchema.shape,
    async (params) => {
      const result = await transformImage(client, params);
      return { content: result };
    }
  );

• src/index.ts:22-25 (registration)

  Import statement for the transform_image tool components (schema, description, handler).

    transformImageSchema,
    transformImageDescription,
    transformImage,
  } from "./tools/transform-image.js";