add URL translation for content extraction, fetch raw over DOM scraping

for known dynamic pages (GitLab CI jobs), fetch the raw/plain text
endpoint instead of scraping a half-loaded DOM. falls back to DOM
extraction with a MutationObserver settle wait when raw isn't accessible.
extractWebpageContent is now async, all callers updated.

public/src/content.js

@@ -1,7 +1,73 @@
 // Content extraction and processing functionality
 
-// Function to extract text content along with links and metadata from the webpage
-export function extractWebpageContent() {
+// URL translation rules — map dynamic page patterns to raw/plain text equivalents.
+// Each rule: { pattern: RegExp matching the full URL, translate: (match) => rawURL }
+const URL_TRANSLATIONS = [
+	{
+		// GitLab CI job pages → raw log output
+		// e.g. https://git.example.com/group/project/-/jobs/12345 → .../jobs/12345/raw
+		pattern: /^(https?:\/\/[^/]+\/.*\/-\/jobs\/\d+)\/?$/,
+		translate: (match) => `${match[1]}/raw`,
+	},
+];
+
+// Try to fetch raw content via URL translation (same-origin, cookies included).
+// Returns { raw: string } on success, { matched: true } if pattern matched but
+// fetch failed (signals a dynamic page worth waiting for), or null if no match.
+async function fetchTranslatedContent(url) {
+	for (const rule of URL_TRANSLATIONS) {
+		const match = url.match(rule.pattern);
+		if (match) {
+			try {
+				const rawUrl = rule.translate(match);
+				const response = await fetch(rawUrl, { credentials: "same-origin" });
+				if (response.ok) {
+					const text = await response.text();
+					if (text && text.length > 0) {
+						return { raw: text };
+					}
+				}
+			} catch (e) {
+				console.warn("URL translation fetch failed, falling back to DOM:", e);
+			}
+			// Pattern matched but fetch failed — caller should wait for DOM to settle
+			return { matched: true };
+		}
+	}
+	return null;
+}
+
+// Wait for DOM mutations to settle (content stops streaming in)
+function waitForDOMSettle(timeout = 3000, quiet = 500) {
+	return new Promise((resolve) => {
+		let timer;
+		const observer = new MutationObserver(() => {
+			clearTimeout(timer);
+			timer = setTimeout(() => {
+				observer.disconnect();
+				resolve();
+			}, quiet);
+		});
+		observer.observe(document.body, {
+			childList: true,
+			subtree: true,
+			characterData: true,
+		});
+		// Start the quiet timer immediately in case DOM is already settled
+		timer = setTimeout(() => {
+			observer.disconnect();
+			resolve();
+		}, quiet);
+		// Hard ceiling so we don't wait forever
+		setTimeout(() => {
+			observer.disconnect();
+			resolve();
+		}, timeout);
+	});
+}
+
+// Extract DOM content — title, metadata, text with links
+function extractDOMContent() {
 	let content = "";
 
 	// Extract title
@@ -40,3 +106,30 @@ export function extractWebpageContent() {
 	getTextWithLinks(document.body);
 	return content.trim();
 }
+
+// Extract text content along with links and metadata from the webpage.
+// Tries URL translation first (raw/plain text for known dynamic pages),
+// falls back to DOM extraction.
+export async function extractWebpageContent() {
+	const result = await fetchTranslatedContent(window.location.href);
+
+	// Raw content fetched successfully — use it directly
+	if (result?.raw) {
+		let content = "";
+		const title = document.title;
+		if (title) {
+			content += `**Page Title**: ${title}\n\n`;
+		}
+		content += `**Source URL**: ${window.location.href}\n\n`;
+		content += result.raw;
+		return content.trim();
+	}
+
+	// Pattern matched but raw fetch failed (auth, CORS, etc) —
+	// this is a known dynamic page, wait for DOM to settle before scraping
+	if (result?.matched) {
+		await waitForDOMSettle();
+	}
+
+	return extractDOMContent();
+}

public/src/page-reader.js

@@ -5,7 +5,7 @@ import { getUIText } from "./ui-translations.js";
 
 // Function to read the entire page using Hermes
 export async function readPageWithHermes(button = null) {
-	const content = extractWebpageContent();
+	const content = await extractWebpageContent();
 
 	// Update button tooltip instead of creating separate DOM element
 	if (button) {

public/src/ui.js

@@ -295,14 +295,14 @@ export function initializeUncloseaiElements() {
 // Widget creation functions imported from widget-library.js
 
 // Initialize the legacy chat interface (for backward compatibility)
-export function initializeChatInterface() {
+export async function initializeChatInterface() {
 	// Only initialize if there are legacy elements (chat-container, user-input, etc.)
 	const legacyElements = document.querySelector(
 		"#chat-container, #user-input, #chat-box",
 	);
 	if (!legacyElements) return;
 
-	const pageContent = extractWebpageContent();
+	const pageContent = await extractWebpageContent();
 	chatHistory.push({
 		role: "system",
 		content: `Here's the content of the webpage: ${pageContent}`,

public/src/uncloseai-embed-modal.js

@@ -1177,7 +1177,7 @@ async function openUncloseaiEmbeddedModalNew() {
 					try {
 						// Get page content
 						const { extractWebpageContent } = await import("./content.js");
-						const pageContent = extractWebpageContent();
+						const pageContent = await extractWebpageContent();
 
 						// Generate TTS
 						const { speakText } = await import("./tts.js");
@@ -1895,7 +1895,7 @@ async function openUncloseaiEmbeddedModalNew() {
 				console.log("History length:", history ? history.length : 0);
 				if (history && history.length > 0) {
 					// Set up page context for existing conversation
-					const pageContent = extractWebpageContent();
+					const pageContent = await extractWebpageContent();
 					const pageTitle = document.title || window.location.hostname;
 					const conversationContextAppend = `
 
@@ -2324,7 +2324,7 @@ You have complete knowledge of this page content and can reference any details,
 
 			try {
 				// Get page content for context
-				const pageContent = extractWebpageContent();
+				const pageContent = await extractWebpageContent();
 				const pageTitle = document.title || window.location.hostname;
 
 				// FIRST: Generate intro message with specialized intro system prompt