Allow selection of multiple browser viewport sizes and adjusting screenshot quality

Author: mrubens

.changeset/five-gorillas-exist.md

@@ -0,0 +1,5 @@
+---
+"roo-cline": patch
+---
+
+Allow selection of multiple browser viewport sizes and adjusting screenshot quality

README.md

@@ -7,7 +7,7 @@ A fork of Cline, an autonomous coding agent, with some additional experimental f
 - Drag and drop images into chats
 - "Enhance prompt" button (OpenRouter models only for now)
 - Sound effects for feedback
-- Option to use a larger 1280x800 browser
+- Option to use browsers of different sizes and adjust screenshot quality
 - Quick prompt copying from history
 - OpenRouter compression support
 - Includes current time in the system prompt

jest.config.js

@@ -29,6 +29,9 @@ module.exports = {
     transformIgnorePatterns: [
         'node_modules/(?!(@modelcontextprotocol|delay|p-wait-for|globby|serialize-error|strip-ansi|default-shell|os-name)/)'
     ],
+    modulePathIgnorePatterns: [
+        '.vscode-test'
+    ],
     setupFiles: [],
     globals: {
         'ts-jest': {

src/core/Cline.ts

@@ -786,8 +786,8 @@ export class Cline {
 			throw new Error("MCP hub not available")
 		}
 
-		const { browserLargeViewport, preferredLanguage } = await this.providerRef.deref()?.getState() ?? {}
-		const systemPrompt = await SYSTEM_PROMPT(cwd, this.api.getModel().info.supportsComputerUse ?? false, mcpHub, this.diffStrategy, browserLargeViewport) + await addCustomInstructions(this.customInstructions ?? '', cwd, preferredLanguage)
+		const { browserViewportSize, preferredLanguage } = await this.providerRef.deref()?.getState() ?? {}
+		const systemPrompt = await SYSTEM_PROMPT(cwd, this.api.getModel().info.supportsComputerUse ?? false, mcpHub, this.diffStrategy, browserViewportSize) + await addCustomInstructions(this.customInstructions ?? '', cwd, preferredLanguage)
 
 		// If the previous API request's total token usage is close to the context window, truncate the conversation history to free up space for the new request
 		if (previousApiReqIndex >= 0) {

src/core/prompts/system.ts

@@ -11,7 +11,7 @@ export const SYSTEM_PROMPT = async (
 	supportsComputerUse: boolean,
 	mcpHub: McpHub,
 	diffStrategy?: DiffStrategy,
-	browserLargeViewport?: boolean
+	browserViewportSize?: string
 ) => `You are Cline, a highly skilled software engineer with extensive knowledge in many programming languages, frameworks, design patterns, and best practices.
 
 ====
@@ -114,7 +114,7 @@ Usage:
 Description: Request to interact with a Puppeteer-controlled browser. Every action, except \`close\`, will be responded to with a screenshot of the browser's current state, along with any new console logs. You may only perform one browser action per message, and wait for the user's response including a screenshot and logs to determine the next action.
 - The sequence of actions **must always start with** launching the browser at a URL, and **must always end with** closing the browser. If you need to visit a new URL that is not possible to navigate to from the current webpage, you must first close the browser, then launch again at the new URL.
 - While the browser is active, only the \`browser_action\` tool can be used. No other tools should be called during this time. You may proceed to use other tools only after closing the browser. For example if you run into an error and need to fix a file, you must close the browser, then use other tools to make the necessary changes, then re-launch the browser to verify the result.
-- The browser window has a resolution of **${browserLargeViewport ? "1280x800" : "900x600"}** pixels. When performing any click actions, ensure the coordinates are within this resolution range.
+- The browser window has a resolution of **${browserViewportSize || "900x600"}** pixels. When performing any click actions, ensure the coordinates are within this resolution range.
 - Before clicking on any elements such as icons, links, or buttons, you must consult the provided screenshot of the page to determine the coordinates of the element. The click should be targeted at the **center of the element**, not on its edges.
 Parameters:
 - action: (required) The action to perform. The available actions are:
@@ -132,7 +132,7 @@ Parameters:
         - Example: \`<action>close</action>\`
 - url: (optional) Use this for providing the URL for the \`launch\` action.
     * Example: <url>https://example.com</url>
-- coordinate: (optional) The X and Y coordinates for the \`click\` action. Coordinates should be within the **${browserLargeViewport ? "1280x800" : "900x600"}** resolution.
+- coordinate: (optional) The X and Y coordinates for the \`click\` action. Coordinates should be within the **${browserViewportSize || "900x600"}** resolution.
     * Example: <coordinate>450,300</coordinate>
 - text: (optional) Use this for providing the text for the \`type\` action.
     * Example: <text>Hello, world!</text>

src/core/webview/ClineProvider.ts

@@ -71,7 +71,8 @@ type GlobalStateKey =
 	| "soundVolume"
 	| "diffEnabled"
 	| "alwaysAllowMcp"
-	| "browserLargeViewport"
+	| "browserViewportSize"
+	| "screenshotQuality"
 	| "fuzzyMatchThreshold"
 	| "preferredLanguage" // Language setting for Cline's communication
 	| "writeDelayMs"
@@ -624,9 +625,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 						await this.updateGlobalState("diffEnabled", diffEnabled)
 						await this.postStateToWebview()
 						break
-					case "browserLargeViewport":
-						const browserLargeViewport = message.bool ?? false
-						await this.updateGlobalState("browserLargeViewport", browserLargeViewport)
+					case "browserViewportSize":
+						const browserViewportSize = message.text ?? "900x600"
+						await this.updateGlobalState("browserViewportSize", browserViewportSize)
 						await this.postStateToWebview()
 						break
 					case "fuzzyMatchThreshold":
@@ -641,6 +642,10 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 						await this.updateGlobalState("writeDelayMs", message.value)
 						await this.postStateToWebview()
 						break
+					case "screenshotQuality":
+						await this.updateGlobalState("screenshotQuality", message.value)
+						await this.postStateToWebview()
+						break
 					case "enhancePrompt":
 						if (message.text) {
 							try {
@@ -1015,7 +1020,8 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			diffEnabled,
 			taskHistory,
 			soundVolume,
-			browserLargeViewport,
+			browserViewportSize,
+			screenshotQuality,
 			preferredLanguage,
 			writeDelayMs,
 		} = await this.getState()
@@ -1043,7 +1049,8 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			shouldShowAnnouncement: lastShownAnnouncementId !== this.latestAnnouncementId,
 			allowedCommands,
 			soundVolume: soundVolume ?? 0.5,
-			browserLargeViewport: browserLargeViewport ?? false,
+			browserViewportSize: browserViewportSize ?? "900x600",
+			screenshotQuality: screenshotQuality ?? 75,
 			preferredLanguage: preferredLanguage ?? 'English',
 			writeDelayMs: writeDelayMs ?? 1000,
 		}
@@ -1140,10 +1147,11 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			soundEnabled,
 			diffEnabled,
 			soundVolume,
-			browserLargeViewport,
+			browserViewportSize,
 			fuzzyMatchThreshold,
 			preferredLanguage,
 			writeDelayMs,
+			screenshotQuality,
 		] = await Promise.all([
 			this.getGlobalState("apiProvider") as Promise<ApiProvider | undefined>,
 			this.getGlobalState("apiModelId") as Promise<string | undefined>,
@@ -1183,10 +1191,11 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			this.getGlobalState("soundEnabled") as Promise<boolean | undefined>,
 			this.getGlobalState("diffEnabled") as Promise<boolean | undefined>,
 			this.getGlobalState("soundVolume") as Promise<number | undefined>,
-			this.getGlobalState("browserLargeViewport") as Promise<boolean | undefined>,
+			this.getGlobalState("browserViewportSize") as Promise<string | undefined>,
 			this.getGlobalState("fuzzyMatchThreshold") as Promise<number | undefined>,
 			this.getGlobalState("preferredLanguage") as Promise<string | undefined>,
 			this.getGlobalState("writeDelayMs") as Promise<number | undefined>,
+			this.getGlobalState("screenshotQuality") as Promise<number | undefined>,
 		])
 
 		let apiProvider: ApiProvider
@@ -1244,7 +1253,8 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			soundEnabled: soundEnabled ?? false,
 			diffEnabled: diffEnabled ?? true,
 			soundVolume,
-			browserLargeViewport: browserLargeViewport ?? false,
+			browserViewportSize: browserViewportSize ?? "900x600",
+			screenshotQuality: screenshotQuality ?? 75,
 			fuzzyMatchThreshold: fuzzyMatchThreshold ?? 1.0,
 			writeDelayMs: writeDelayMs ?? 1000,
 			preferredLanguage: preferredLanguage ?? (() => {

src/core/webview/__tests__/ClineProvider.test.ts

@@ -253,7 +253,7 @@ describe('ClineProvider', () => {
             soundEnabled: false,
             diffEnabled: false,
             writeDelayMs: 1000,
-            browserLargeViewport: false,
+            browserViewportSize: "900x600",
             fuzzyMatchThreshold: 1.0,
         }
 

src/services/browser/BrowserSession.ts

@@ -58,9 +58,11 @@ export class BrowserSession {
 				"--user-agent=Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36",
 			],
 			executablePath: stats.executablePath,
-			defaultViewport: await this.context.globalState.get("browserLargeViewport")
-				? { width: 1280, height: 800 }
-				: { width: 900, height: 600 },
+			defaultViewport: (() => {
+				const size = (this.context.globalState.get("browserViewportSize") as string | undefined) || "900x600"
+				const [width, height] = size.split("x").map(Number)
+				return { width, height }
+			})(),
 			// headless: false,
 		})
 		// (latest version of puppeteer does not add headless to user agent)
@@ -134,7 +136,7 @@ export class BrowserSession {
 		let screenshotBase64 = await this.page.screenshot({
 			...options,
 			type: "webp",
-			quality: 100, // Set maximum quality to prevent compression artifacts
+			quality: (await this.context.globalState.get("screenshotQuality") as number | undefined) ?? 75,
 		})
 		let screenshot = `data:image/webp;base64,${screenshotBase64}`
 
@@ -245,27 +247,29 @@ export class BrowserSession {
 	}
 
 	async scrollDown(): Promise<BrowserActionResult> {
-		const isLargeViewport = await this.context.globalState.get("browserLargeViewport")
+		const size = (await this.context.globalState.get("browserViewportSize") as string | undefined) || "900x600"
+		const height = parseInt(size.split("x")[1])
 		return this.doAction(async (page) => {
 			await page.evaluate((scrollHeight) => {
 				window.scrollBy({
 					top: scrollHeight,
 					behavior: "auto",
 				})
-			}, isLargeViewport ? 800 : 600)
+			}, height)
 			await delay(300)
 		})
 	}
 
 	async scrollUp(): Promise<BrowserActionResult> {
-		const isLargeViewport = await this.context.globalState.get("browserLargeViewport")
+		const size = (await this.context.globalState.get("browserViewportSize") as string | undefined) || "900x600"
+		const height = parseInt(size.split("x")[1])
 		return this.doAction(async (page) => {
 			await page.evaluate((scrollHeight) => {
 				window.scrollBy({
 					top: -scrollHeight,
 					behavior: "auto",
 				})
-			}, isLargeViewport ? 800 : 600)
+			}, height)
 			await delay(300)
 		})
 	}

src/shared/ExtensionMessage.ts

@@ -56,7 +56,8 @@ export interface ExtensionState {
 	soundEnabled?: boolean
 	soundVolume?: number
 	diffEnabled?: boolean
-	browserLargeViewport?: boolean
+	browserViewportSize?: string
+	screenshotQuality?: number
 	fuzzyMatchThreshold?: number
 	preferredLanguage: string
 	writeDelayMs: number

src/shared/WebviewMessage.ts

@@ -35,7 +35,8 @@ export interface WebviewMessage {
 		| "soundEnabled"
 		| "soundVolume"
 		| "diffEnabled"
-		| "browserLargeViewport"
+		| "browserViewportSize"
+		| "screenshotQuality"
 		| "openMcpSettings"
 		| "restartMcpServer"
 		| "toggleToolAlwaysAllow"