better title gen & conv ordering

src/lib/backend/convex/conversations.ts

@@ -28,6 +28,7 @@ export const get = query({
 			.collect();
 
 		return conversations.sort((a, b) => {
+			if (a.generating && b.generating) return 0;
 			const aTime = a.updated_at ?? 0;
 			const bTime = b.updated_at ?? 0;
 

src/routes/api/generate-message/+server.ts

@@ -108,18 +108,23 @@ async function generateConversationTitle({
 	});
 
 	// Create a prompt for title generation using only the first user message
-	const titlePrompt = `Based on this user request, generate a concise, specific title (max 4-5 words):
+	const titlePrompt = `Based on this message:
+"""${userMessage}"""
 
-${userMessage}
+Generate a concise, specific title (max 4-5 words).
+Generate only the title based on the message, nothing else. Don't name the title 'Generate Title' or anything stupid like that, otherwise its obvious we're generating a title with AI.
 
-Generate only the title based on what the user is asking for, nothing else:`;
+Also, do not interact with the message directly or answer it. Just generate the title based on the message.
+
+If its a simple hi, just name it "Greeting" or something like that.
+`;
 
 	const titleResult = await ResultAsync.fromPromise(
 		openai.chat.completions.create({
 			model: 'mistralai/ministral-8b',
 			messages: [{ role: 'user', content: titlePrompt }],
 			max_tokens: 20,
-			temperature: 0.3,
+			temperature: 0.5,
 		}),
 		(e) => `Title generation API call failed: ${e}`
 	);
@@ -149,6 +154,7 @@ Generate only the title based on what the user is asking for, nothing else:`;
 		}),
 		(e) => `Failed to update conversation title: ${e}`
 	);
+	t;
 
 	if (updateResult.isErr()) {
 		log(`Title generation: Failed to update title: ${updateResult.error}`, startTime);
@@ -184,19 +190,20 @@ async function generateAIResponse({
 		return;
 	}
 
-	const [modelResult, keyResult, messagesQueryResult, rulesResult, userSettingsResult] = await Promise.all([
-		modelResultPromise,
-		keyResultPromise,
-		ResultAsync.fromPromise(
-			client.query(api.messages.getAllFromConversation, {
-				conversation_id: conversationId as Id<'conversations'>,
-				session_token: sessionToken,
-			}),
-			(e) => `Failed to get messages: ${e}`
-		),
-		rulesResultPromise,
-		userSettingsPromise,
-	]);
+	const [modelResult, keyResult, messagesQueryResult, rulesResult, userSettingsResult] =
+		await Promise.all([
+			modelResultPromise,
+			keyResultPromise,
+			ResultAsync.fromPromise(
+				client.query(api.messages.getAllFromConversation, {
+					conversation_id: conversationId as Id<'conversations'>,
+					session_token: sessionToken,
+				}),
+				(e) => `Failed to get messages: ${e}`
+			),
+			rulesResultPromise,
+			userSettingsPromise,
+		]);
 
 	if (modelResult.isErr()) {
 		handleGenerationError({
@@ -311,7 +318,8 @@ async function generateAIResponse({
 
 			if (freeMessagesUsed >= 10) {
 				handleGenerationError({
-					error: 'Free message limit reached (10/10). Please add your own OpenRouter API key to continue chatting, or use a free model ending in ":free".',
+					error:
+						'Free message limit reached (10/10). Please add your own OpenRouter API key to continue chatting, or use a free model ending in ":free".',
 					conversationId,
 					messageId: mid,
 					sessionToken,
@@ -518,12 +526,15 @@ ${attachedRules.map((r) => `- ${r.name}: ${r.rule}`).join('\n')}`,
 			(e) => `Failed to render HTML: ${e}`
 		);
 
-		const generationStatsResult = await retryResult(() => getGenerationStats(generationId!, actualKey), {
-			delay: 500,
-			retries: 2,
-			startTime,
-			fnName: 'getGenerationStats',
-		});
+		const generationStatsResult = await retryResult(
+			() => getGenerationStats(generationId!, actualKey),
+			{
+				delay: 500,
+				retries: 2,
+				startTime,
+				fnName: 'getGenerationStats',
+			}
+		);
 
 		if (generationStatsResult.isErr()) {
 			log(`Background: Failed to get generation stats: ${generationStatsResult.error}`, startTime);

src/routes/chat/search-modal.svelte

@@ -148,9 +148,6 @@
 								<div class={['truncate font-medium', titleMatch && 'text-heading']}>
 									{conversation.title}
 								</div>
-								<div class="text-muted-foreground bg-muted rounded px-1.5 py-0.5 text-xs">
-									{Math.round(score * 100)}%
-								</div>
 							</div>
 							<div class="text-muted-foreground text-xs">
 								{messages.length} matching message{messages.length !== 1 ? 's' : ''}