feat: add optional input image parameter to image generation tool

- Added optional image parameter to generate_image tool
- Updated tool description to document the new parameter
- Modified generateImageTool to read and validate input images
- Updated OpenRouter API to include input image in requests
- Added comprehensive test coverage for new functionality

Author: roomote

src/api/providers/openrouter.ts

@@ -275,9 +275,15 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 	 * @param prompt The text prompt for image generation
 	 * @param model The model to use for generation
 	 * @param apiKey The OpenRouter API key (must be explicitly provided)
+	 * @param inputImage Optional base64 encoded input image data URL
 	 * @returns The generated image data and format, or an error
 	 */
-	async generateImage(prompt: string, model: string, apiKey: string): Promise<ImageGenerationResult> {
+	async generateImage(
+		prompt: string,
+		model: string,
+		apiKey: string,
+		inputImage?: string,
+	): Promise<ImageGenerationResult> {
 		if (!apiKey) {
 			return {
 				success: false,
@@ -299,7 +305,20 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 					messages: [
 						{
 							role: "user",
-							content: prompt,
+							content: inputImage
+								? [
+										{
+											type: "text",
+											text: prompt,
+										},
+										{
+											type: "image_url",
+											image_url: {
+												url: inputImage,
+											},
+										},
+									]
+								: prompt,
 						},
 					],
 					modalities: ["image", "text"],

src/core/prompts/tools/generate-image.ts

@@ -2,19 +2,28 @@ import { ToolArgs } from "./types"
 
 export function getGenerateImageDescription(args: ToolArgs): string {
 	return `## generate_image
-Description: Request to generate an image using AI models through OpenRouter API. This tool creates images from text prompts and saves them to the specified path.
+Description: Request to generate an image using AI models through OpenRouter API. This tool creates images from text prompts and saves them to the specified path. Optionally, you can provide an input image to use as a reference or starting point for the generation.
 Parameters:
 - prompt: (required) The text prompt describing the image to generate
 - path: (required) The file path where the generated image should be saved (relative to the current workspace directory ${args.cwd}). The tool will automatically add the appropriate image extension if not provided.
+- image: (optional) The file path to an input image to use as a reference or starting point (relative to the current workspace directory ${args.cwd}). Supported formats: PNG, JPG, JPEG, GIF, WEBP.
 Usage:
 <generate_image>
 <prompt>Your image description here</prompt>
 <path>path/to/save/image.png</path>
+<image>path/to/input/image.jpg</image>
 </generate_image>
 
 Example: Requesting to generate a sunset image
 <generate_image>
 <prompt>A beautiful sunset over mountains with vibrant orange and purple colors</prompt>
 <path>images/sunset.png</path>
+</generate_image>
+
+Example: Generating an image with an input reference
+<generate_image>
+<prompt>Transform this image into a watercolor painting style</prompt>
+<path>images/watercolor-output.png</path>
+<image>images/original-photo.jpg</image>
 </generate_image>`
 }