feat: in-session conversation memory for AI chat

Passes the full conversation history (up to 20 turns) on each request
so the AI remembers earlier messages within the same session.

Author: Spectual

backend/app.py

@@ -114,6 +114,7 @@ def chat():
     try:
         data = request.get_json()
         message = data.get('message', '')
+        history = data.get('history', [])[-20:]  # cap at 20 messages
 
         if not message:
             return jsonify({'error': 'Message is required'}), 400
@@ -211,12 +212,18 @@ def chat():
 - Always respond in the same language as the user's question and respond as if you are {personal_info['name']}.
 """
 
+        # Build messages with conversation history
+        openai_messages = [{"role": "system", "content": final_answer_prompt}]
+        for h in history:
+            role = h.get('role', 'user')
+            if role in ('user', 'assistant'):
+                openai_messages.append({"role": role, "content": h.get('content', '')})
+        openai_messages.append({"role": "user", "content": message})
+
         # Call OpenAI API for the final answer
         final_response = client.chat.completions.create(
             model="gpt-4o-mini",
-            messages=[
-                {"role": "system", "content": final_answer_prompt}
-            ],
+            messages=openai_messages,
             temperature=0.5,
             max_tokens=1000,
         )
@@ -249,6 +256,7 @@ def chat_stream():
     try:
         data = request.get_json()
         message = data.get('message', '')
+        history = data.get('history', [])[-20:]  # cap at 20 messages
 
         if not message:
             return jsonify({'error': 'Message is required'}), 400
@@ -319,16 +327,25 @@ def chat_stream():
 - Always respond in the same language as the user's question and respond as if you are {personal_info['name']}.
 """
 
-        # Capture user_id in closure before streaming
+        # Build messages with conversation history
+        stream_messages = [{"role": "system", "content": final_answer_prompt}]
+        for h in history:
+            role = h.get('role', 'user')
+            if role in ('user', 'assistant'):
+                stream_messages.append({"role": role, "content": h.get('content', '')})
+        stream_messages.append({"role": "user", "content": message})
+
+        # Capture variables in closure before streaming
         captured_user_id = user_id
         captured_message = message
+        captured_stream_messages = stream_messages
 
         def generate():
             full_response = []
             try:
                 stream = client.chat.completions.create(
                     model="gpt-4o-mini",
-                    messages=[{"role": "system", "content": final_answer_prompt}],
+                    messages=captured_stream_messages,
                     temperature=0.5,
                     max_tokens=1000,
                     stream=True,

src/components/ChatSection.tsx

@@ -1,7 +1,7 @@
 import { useState, useEffect, useRef } from "react";
 import { AlertCircle, Download } from "lucide-react";
 import MessageBubble from "./MessageBubble";
-import { sendMessageStream, checkHealth } from "@/utils/api";
+import { sendMessageStream, checkHealth, type ConversationMessage } from "@/utils/api";
 import { toast } from "sonner";
 import type { Message } from "@/types/chat";
 import { useLang } from "@/i18n/LanguageContext";
@@ -152,6 +152,15 @@ const ChatSection = () => {
       timestamp: new Date(),
     };
 
+    // Build conversation history for the backend (exclude initial greeting, cap at 20 messages)
+    const history: ConversationMessage[] = messages
+      .filter((m) => m.id !== INITIAL_MSG_ID)
+      .slice(-20)
+      .map((m) => ({
+        role: m.isUser ? 'user' : 'assistant',
+        content: m.text,
+      }));
+
     setMessages((prev) => [...prev, userMessage]);
     setInputValue("");
     setIsLoading(true);
@@ -161,6 +170,7 @@ const ChatSection = () => {
 
     await sendMessageStream(
       messageText,
+      history,
       // onChunk: first chunk creates the message, subsequent ones append
       (chunk) => {
         setIsLoading(false);

src/utils/api.ts

@@ -7,14 +7,19 @@ export interface ChatResponse {
   error?: string;
 }
 
-export async function sendMessage(message: string): Promise<ChatResponse> {
+export interface ConversationMessage {
+  role: 'user' | 'assistant';
+  content: string;
+}
+
+export async function sendMessage(message: string, history: ConversationMessage[] = []): Promise<ChatResponse> {
   try {
     const response = await fetch(`${API_BASE_URL}/api/chat`, {
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
       },
-      body: JSON.stringify({ message }),
+      body: JSON.stringify({ message, history }),
     });
 
     if (!response.ok) {
@@ -35,6 +40,7 @@ export async function sendMessage(message: string): Promise<ChatResponse> {
 
 export async function sendMessageStream(
   message: string,
+  history: ConversationMessage[],
   onChunk: (chunk: string) => void,
   onDone: () => void,
   onError: (error: string) => void
@@ -43,7 +49,7 @@ export async function sendMessageStream(
     const response = await fetch(`${API_BASE_URL}/api/chat/stream`, {
       method: 'POST',
       headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({ message }),
+      body: JSON.stringify({ message, history }),
     });
 
     if (!response.ok) {