chrome-ai-bridge 1.0.14 → 1.0.16

package/build/src/tools/gemini-web.js

@@ -267,41 +267,31 @@ export const askGeminiWeb = defineTool({
             }
             response.appendResponseLine('✅ ログイン確認完了');
             response.appendResponseLine('質問を送信中...');
-            // Input text using the textbox element
-            // Gemini uses a textbox with role="textbox" or a div with contenteditable
-            // NOTE: Gemini has Trusted Types CSP, so we cannot use innerHTML directly
-            const questionSent = await page.evaluate(questionText => {
-                // Helper to clear element content without innerHTML (CSP-safe)
-                const clearElement = (el) => {
-                    while (el.firstChild) {
-                        el.removeChild(el.firstChild);
-                    }
-                };
-                // Try textbox first (Gemini's current implementation)
-                const textbox = document.querySelector('[role="textbox"]');
-                if (textbox) {
-                    textbox.focus();
-                    // Clear existing content (CSP-safe)
-                    clearElement(textbox);
-                    // Insert text using textContent (CSP-safe)
-                    textbox.textContent = questionText;
-                    textbox.dispatchEvent(new Event('input', { bubbles: true }));
-                    return true;
-                }
-                // Fallback to contenteditable
-                const editor = document.querySelector('div[contenteditable="true"]');
-                if (editor) {
-                    clearElement(editor);
-                    editor.textContent = questionText;
-                    editor.dispatchEvent(new Event('input', { bubbles: true }));
-                    return true;
-                }
-                return false;
-            }, sanitizedQuestion);
-            if (!questionSent) {
+            // Input text using Puppeteer's type() for proper Angular state updates
+            // Gemini uses Angular and requires real keyboard events to update internal state
+            const textboxSelector = '[role="textbox"]';
+            const textbox = await page.$(textboxSelector);
+            if (!textbox) {
                 response.appendResponseLine('❌ 入力欄が見つかりません');
                 return;
             }
+            // Click to focus and clear existing content
+            await textbox.click({ clickCount: 3 }); // Triple-click to select all
+            await page.keyboard.press('Backspace'); // Clear selection
+            // Type the question using real keyboard events (essential for Angular)
+            // IMPORTANT: Split by newlines and use Shift+Enter for line breaks
+            // (Enter alone triggers send in Gemini)
+            const lines = sanitizedQuestion.split('\n');
+            for (let i = 0; i < lines.length; i++) {
+                if (i > 0) {
+                    // Shift+Enter for newline (Enter alone sends the message)
+                    await page.keyboard.down('Shift');
+                    await page.keyboard.press('Enter');
+                    await page.keyboard.up('Shift');
+                }
+                await page.keyboard.type(lines[i], { delay: 2 });
+            }
+            // Wait for Angular to process the input and show send button
             await new Promise(resolve => setTimeout(resolve, 500));
             // 質問送信前に model-response 要素数を記録（ChatGPTと同じカウント方式）
             const initialModelResponseCount = await page.evaluate(() => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "chrome-ai-bridge",
-  "version": "1.0.14",
+  "version": "1.0.16",
   "description": "MCP server bridging Chrome browser and AI assistants (ChatGPT, Gemini). Browser automation + AI consultation.",
   "type": "module",
   "bin": "./scripts/cli.mjs",