Mirror of github.com/Abhisingh18/Trust-first-AI-Copilot

README.md

@@ -0,0 +1,225 @@
+# Trust-First AI Copilot
+**Perplexity-Style • System-Driven • No Custom LLM**
+Deployed Link: https://trust-first-ai.vercel.app/
+
+A trust-first AI Copilot that delivers **verified, source-grounded, and confidence-scored answers** using strict system rules — inspired by Perplexity and designed to fix the core limitations of modern AI copilots.
+
+---
+
+## Problem
+Most AI copilots today:
+- Produce confident but incorrect (hallucinated) answers
+- Lose context in long or multi-file documents
+- Hide sources and assumptions
+- Provide limited admin visibility and control
+- Encourage blind dependency on AI outputs
+
+These issues lead to **wrong decisions, rework, and low trust**.
+
+---
+
+## Solution
+This project implements a **system-first AI Copilot** where:
+- Retrieval is mandatory (no context → no answer)
+- Every answer is backed by sources
+- Confidence is explicitly shown
+- Low-confidence answers are refused
+- Automation is human-approved
+- Security, transparency, and auditability are built-in
+
+The focus is **system design over model size**.
+
+---
+
+## Core Principles
+- No guessing
+- No hidden sources
+- No blind automation
+- Refusal is a feature, not a failure
+
+---
+
+## Key Features
+- Mandatory Retrieval-Augmented Generation (RAG)
+- Source-linked answers with citations
+- Confidence scoring (High / Medium / Low)
+- Automatic refusal on insufficient data
+- Workspace / project-level context memory
+- Intent detection and auto-clarification
+- Human-in-the-loop automation (n8n)
+- Zero-trust data access
+- Full audit logs (OpenTelemetry)
+- Model-agnostic LLM layer (Groq)
+
+---
+
+## System Architecture
+User
+→ Intent Detection
+→ Search & Retrieval (Tavily + Vector DB)
+→ Context Ranking & Filtering
+→ LLM (Groq – language & reasoning only)
+→ Verification & Confidence Engine
+→ Answer + Sources + Assumptions
+→ (Optional) Human-Approved Automation
+→ Audit Logs & Admin Dashboard
+
+
+## What This Project Is Not
+- Not a chatbot
+- Not prompt-dependent
+- Not blind AI
+- Not a Copilot replacement
+
+This is a **controlled, transparent, enterprise-ready AI system**.
+
+---
+
+## Tech Stack
+**Frontend**
+- Next.js
+- React
+- Tailwind CSS
+
+**Backend**
+- FastAPI (Python)
+
+**AI & Data**
+- LLM: Groq (LLaMA / Mixtral)
+- Search: Tavily API
+- Embeddings: Hugging Face / Local models
+- Vector DB: FAISS / Qdrant
+- Automation: n8n
+- Logging & Audit: OpenTelemetry
+
+**Deployment**
+- Frontend: Vercel
+- Backend: Render
+
+---
+
+## Project Structure
+project-root/
+├── frontend/
+│ ├── pages/
+│ ├── components/
+│ └── services/
+├── backend/
+│ ├── main.py
+│ ├── rag/
+│ ├── verification/
+│ ├── automation/
+│ └── requirements.txt
+├── docs/
+└── README.md
+
+yaml
+Copy code
+
+---
+
+## Required API Keys
+| Service | Purpose |
+|-------|---------|
+| Groq | LLM inference |
+| Tavily | Web search |
+| Hugging Face | Embeddings |
+| n8n | Automation |
+
+> All API keys are stored **only in backend environment variables**.
+
+---
+
+## Local Setup (Backend)
+```bash
+git clone https://github.com/your-username/your-repo.git
+cd backend
+python -m venv venv
+source venv/bin/activate
+pip install -r requirements.txt
+Create .env:
+
+env
+Copy code
+GROQ_API_KEY=xxxx
+TAVILY_API_KEY=tvly_xxxx
+HF_API_KEY=hf_xxxx
+N8N_API_KEY=xxxx
+N8N_BASE_URL=http://localhost:5678
+Run backend:
+
+bash
+Copy code
+uvicorn main:app --reload
+Local Setup (Frontend)
+bash
+Copy code
+cd frontend
+npm install
+npm run dev
+Deployment
+Backend
+
+Push code to GitHub
+
+Connect repository to Render
+
+Build command:
+
+bash
+Copy code
+pip install -r requirements.txt
+Start command:
+
+bash
+Copy code
+uvicorn main:app --host 0.0.0.0 --port 10000
+Frontend
+
+Deploy via Vercel
+
+Set backend API URL in environment variables
+
+Security & Trust Model
+API keys never exposed to frontend
+
+Per-user data isolation
+
+Role-based access control
+
+Full audit trail for AI actions
+
+How Hallucinations Are Prevented
+Retrieval is mandatory
+
+Claims must map to sources
+
+Confidence is evaluated
+
+Low confidence triggers refusal
+
+No source → No answer
+
+Use Cases
+Research and academic assistance
+
+Enterprise internal knowledge copilots
+
+Policy and compliance analysis
+
+Long-document summarization
+
+Decision-support systems
+
+Future Improvements
+Offline read-only mode
+
+Multimodal reasoning (charts + text)
+
+Advanced admin dashboards
+
+Domain-specific copilots
+
+Final Note
+LLMs don’t fail — systems fail.
+This project demonstrates how strong system design beats larger models.

app/api/endpoints.py

@@ -0,0 +1,182 @@
+from fastapi import APIRouter, HTTPException
+from pydantic import BaseModel
+from typing import List, Optional
+from app.services.llm import llm_service
+from app.services.vector import vector_service
+from app.services.search import search_service
+from app.services.intent import IntentService
+from app.services.files import file_service
+from fastapi import UploadFile, File
+
+# Initialize Intent Service
+intent_service = IntentService(llm_service)
+
+router = APIRouter()
+
+# --- Pydantic Models ---
+class QueryRequest(BaseModel):
+    query: str
+
+class Source(BaseModel):
+    title: str
+    url: str
+    snippet: str
+
+class ChallengeRequest(BaseModel):
+    original_query: str
+    original_answer: str
+    sources_text: str
+
+class QueryResponse(BaseModel):
+    answer: str
+    sources: List[Source]
+    confidence: str
+    search_queries: List[str]
+    intent: Optional[str] = None
+    thought_process: Optional[str] = None
+
+# --- Endpoints ---
+
+@router.post("/query", response_model=QueryResponse)
+async def process_query(request: QueryRequest):
+    """
+    Main orchestration endpoint for the Trust-First Copilot.
+    """
+    user_query = request.query
+    print(f"Refining query: {user_query}")
+
+    try:
+        # --- PHASE 1: INTENT & RISK ANALYSIS ---
+        print("🧠 Analyzing Intent...")
+        try:
+            intent = await intent_service.analyze(user_query)
+            print(f"   Category: {intent.category}")
+            print(f"   Reasoning: {intent.reasoning}")
+            print(f"   Risk: {intent.risk_level}")
+        except Exception as e:
+            print(f"Intent Error: {e}")
+            from app.services.intent import IntentResponse
+            intent = IntentResponse(category="SEARCH_REQUIRED", reasoning="Error", risk_level="LOW")
+
+        # Risk Guard
+        if intent.risk_level == "HIGH":
+            return QueryResponse(
+                answer="I cannot fulfill this request as it has been flagged as high risk/safety violation.",
+                sources=[],
+                confidence="Blocked",
+                search_queries=[],
+                intent="High Risk",
+                thought_process=f"Blocked by Risk Analyzer. Reasoning: {intent.reasoning}"
+            )
+
+        # --- PHASE 2: EXECUTION ---
+        search_results = []
+        
+        # Branch 1: Needs Search
+        if intent.category == "SEARCH_REQUIRED" or intent.category == "DATA_ANALYSIS":
+             print("🔍 Initiating Web Search...")
+             search_results = await search_service.search(user_query)
+             if not search_results:
+                 # Fallback if search finds nothing but intent was search
+                 pass
+
+        # Branch 2: Coding (Skip Search usually, unless specific docs needed)
+        elif intent.category == "CODING_TASK":
+            print("💻 Coding Task - Focused Generation")
+            # Potential future improvement: Search for docs if needed
+        
+        # Branch 3: Chat / General
+        else:
+            print("💬 Chat Mode - Direct Generation")
+
+        # --- PHASE 3: CONTEXT & RAG ---
+        context_text = ""
+        final_sources = []
+        
+        if search_results:
+            # RAG Logic
+            print("Indexing search results in Vector DB...")
+            vector_service.create_index_from_results(search_results)
+            
+            print("Searching Vector DB for relevant context...")
+            relevant_chunks = vector_service.search_similar(user_query, k=5)
+            final_sources = relevant_chunks if relevant_chunks else search_results
+            
+            context_text = "\n\n".join([
+                f"Source {i+1}:\nTitle: {r.get('title')}\nURL: {r.get('url')}\nContent: {r.get('content')}"
+                for i, r in enumerate(final_sources)
+            ])
+        else:
+            context_text = "No external sources used. Answering from internal knowledge."
+
+        # --- PHASE 4: SYNTHESIS ---
+        # Modify prompt based on intent? For now, standard synthesis but context aware.
+        answer = await llm_service.synthesize_answer(user_query, context_text)
+
+        # --- PHASE 5: VERIFICATION ---
+        confidence_level = "Medium"
+        if intent.category == "SEARCH_REQUIRED":
+             confidence_assessment = await llm_service.verify_confidence(answer, context_text)
+             if "High confidence" in confidence_assessment: confidence_level = "High"
+             elif "Low confidence" in confidence_assessment: confidence_level = "Low"
+        else:
+            confidence_level = "N/A (Chat)"
+
+        # Construct Response
+        formatted_sources = [
+            Source(title=r.get('title', 'Unknown'), url=r.get('url', '#'), snippet=r.get('content', '')[:200])
+            for r in (search_results if search_results else [])
+        ]
+
+        return QueryResponse(
+            answer=answer,
+            sources=formatted_sources,
+            confidence=confidence_level,
+            search_queries=[user_query],
+            intent=intent.category,
+            thought_process=f"Intent: {intent.category}. Reasoning: {intent.reasoning}"
+        )
+
+    except Exception as e:
+        print(f"Error processing query: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+@router.post("/challenge", response_model=QueryResponse)
+async def challenge_answer(request: ChallengeRequest):
+    """
+    'Disagree-with-Me' Mode: Critiques the previous answer.
+    """
+    try:
+        from app.core import prompts
+        # Construct the critique prompt
+        messages = [
+            {"role": "system", "content": prompts.MASTER_PROMPT_CHALLENGE},
+            {"role": "user", "content": f"Query: {request.original_query}\n\nAnswer to critique: {request.original_answer}\n\nSources used:\n{request.sources_text}"}
+        ]
+        
+        critique = await llm_service._generate(messages, temperature=0.7)
+        
+        # Return as a new message, but marked as a critique
+        return QueryResponse(
+            answer=critique,
+            sources=[],
+            confidence="High (Critique)",
+            search_queries=[],
+            intent="CRITIQUE",
+            thought_process="Devil's Advocate Mode Activated."
+        )
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@router.post("/upload")
+async def upload_file(file: UploadFile = File(...)):
+    """
+    Parses an uploaded file and returns its text content for RAG.
+    """
+    try:
+        filename = file.filename
+        print(f"📂 Processing file: {filename}")
+        content = await file_service.process_file(file)
+        return {"filename": filename, "content": content}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Upload failed: {str(e)}")

app/core/config.py

@@ -0,0 +1,28 @@
+from pydantic_settings import BaseSettings
+
+class Settings(BaseSettings):
+    PROJECT_NAME: str = "Trust-First Copilot"
+    API_V1_STR: str = "/api/v1"
+    
+    # LLM Provider (OpenAI/Anthropic/Azure)
+    OPENAI_API_KEY: str = ""
+    
+    # Search Provider (Tavily/Serper)
+    TAVILY_API_KEY: str = ""
+
+    # Hugging Face Provider
+    HUGGINGFACE_API_KEY: str = ""
+
+    # Groq Provider
+    GROQ_API_KEY: str = ""
+
+    # Gemini Provider
+    GEMINI_API_KEY: str = ""
+
+    # CORS
+    BACKEND_CORS_ORIGINS: list[str] = ["http://localhost:3000"]
+
+    class Config:
+        env_file = ".env"
+
+settings = Settings()

app/core/prompts.py

@@ -0,0 +1,164 @@
+# Master Prompts for Trust-First Copilot
+# These prompts MUST be used verbatim to ensure the "Trust Engine" behavior.
+
+SYSTEM_PROMPT = """
+You are NOT a chatbot.
+
+You are a verification-first, search-grounded AI system similar to Perplexity.
+
+Your job is NOT to guess.
+Your job is to:
+- search
+- retrieve
+- verify
+- summarize
+- cite
+
+STRICT RULES:
+1. If you do not find reliable sources, you MUST say "I don't have enough data to answer this."
+2. Every factual claim MUST be backed by a source.
+3. No source = no answer.
+4. Estimates must clearly state assumptions.
+5. Confidence > verbosity.
+6. Refusal is better than hallucination.
+
+You are building user trust, not impressing users.
+"""
+
+MASTER_PROMPT_SEARCH = """
+You are a search and retrieval engine.
+
+TASK:
+Given a user question, do the following:
+
+1. Break the question into searchable sub-queries
+2. Identify what kind of sources are needed (reports, papers, govt data, official websites)
+3. Retrieve ONLY high-quality, reliable sources
+4. Reject blogs, opinions, or unclear sources
+5. List the sources with:
+   - Source name
+   - Year
+   - Authority (who published it)
+
+DO NOT answer the question yet.
+ONLY return the list of sources and why they are relevant.
+"""
+
+MASTER_PROMPT_ANSWER = """
+You are an answer synthesis engine in 'Truth-First' mode.
+
+INPUT:
+- User question
+- Retrieved sources (from previous step)
+
+RULES:
+1. Answer ONLY using the provided sources.
+2. Separate FACTS from ASSUMPTIONS clearly.
+3. If information is missing, list it under 'Unknowns'.
+
+FORMAT:
+
+### ✅ Verified Facts
+(List only what is explicitly supported by sources)
+
+### ⚠️ Assumptions & Risks
+(List any logical leaps, estimates, or potential inaccuracies)
+
+### ❌ Unknowns / Limitations
+(List what the sources did NOT cover)
+
+### 💡 Synthesis
+(Your cohesive answer. Use the following structure to make it engaging and clear:)
+
+[Direct Answer: 1-2 sentences summarizing the core response]
+
+### **📊 Key Details**
+(Use bullet points. **Bold** the key value or term in each bullet. e.g., "- **Population:** 1.4B")
+
+### **🧑‍🤝‍🧑 Demographic / Structural Breakdown**
+(If applicable, break down by category. **Bold** category names.)
+
+### **🌍 Why This Matters**
+(Explain the significance. **Bold** the main implication.)
+
+### **⚠️ Considerations**
+(Briefly mention major challenges. **Bold** the core challenge.)
+
+Tone:
+Neutral, factual, professional. **Use bolding liberally for emphasis and scannability.**
+"""
+
+MASTER_PROMPT_CHALLENGE = """
+You are a 'Devil's Advocate' Engine.
+
+Your task is to CRITIQUE the following answer.
+Don't be polite. Be rigorous.
+
+Identify:
+1. Logical fallacies.
+2. Unverified assumptions.
+3. Alternative explanations that were ignored.
+
+Output format:
+"Here is why this answer might be wrong..."
+"""
+
+MASTER_PROMPT_CITATION = """
+You are a citation and transparency engine.
+
+TASK:
+For the generated answer:
+1. Map each factual statement to its exact source
+2. Display citations inline or as numbered references
+3. Clearly separate:
+   - Facts
+   - Estimates
+   - Assumptions
+
+OUTPUT MUST INCLUDE:
+- Sources section
+- What is known with high confidence
+- What is uncertain or estimated
+"""
+
+MASTER_PROMPT_CONFIDENCE = """
+You are a confidence evaluation system.
+
+Evaluate the answer and assign:
+- High confidence
+- Medium confidence
+- Low confidence
+
+RULES:
+- High: Fully supported by multiple reliable sources
+- Medium: Partially supported or indirect evidence
+- Low: Weak or missing evidence → REFUSE answer
+
+If refusing:
+- Clearly explain why
+- Suggest what data would be needed to answer
+
+Never bluff.
+Never overstate.
+"""
+
+MASTER_PROMPT_FORMAT = """
+Present the final answer in a website-friendly format.
+
+SECTIONS:
+1. Direct Answer (1–2 lines)
+2. Explanation
+3. Sources (clickable)
+4. Confidence Level
+5. Assumptions / Limitations
+6. Shareable Summary (optional)
+
+DESIGN PRINCIPLES:
+- Clean
+- Minimal
+- Search-first
+- Trust-focused
+
+Avoid chatty language.
+Avoid emojis.
+"""

app/services/files.py

@@ -0,0 +1,63 @@
+import fitz  # PyMuPDF
+import pandas as pd
+from fastapi import UploadFile
+import io
+
+class FileService:
+    async def process_file(self, file: UploadFile) -> str:
+        content_type = file.content_type
+        filename = file.filename.lower()
+        
+        content = await file.read()
+        extracted_text = ""
+
+        try:
+            # PDF Processing
+            if "pdf" in content_type or filename.endswith(".pdf"):
+                extracted_text = self._process_pdf(content)
+            
+            # Excel/CSV Processing
+            elif "excel" in content_type or "spreadsheet" in content_type or filename.endswith((".xlsx", ".xls", ".csv")):
+                extracted_text = self._process_spreadsheet(content, filename)
+            
+            # Text Processing
+            elif "text" in content_type or filename.endswith(".txt") or filename.endswith(".md"):
+                extracted_text = content.decode("utf-8")
+            
+            else:
+                return f"Unsupported file type: {filename}"
+            
+            return extracted_text
+
+        except Exception as e:
+            return f"Error processing file {filename}: {str(e)}"
+
+    def _process_pdf(self, content: bytes) -> str:
+        doc = fitz.open(stream=content, filetype="pdf")
+        text_blocks = []
+        
+        for page_num, page in enumerate(doc):
+            text = page.get_text()
+            if text.strip():
+                # Add page marker for citation
+                text_blocks.append(f"--- Page {page_num + 1} ---\n{text}")
+        
+        return "\n\n".join(text_blocks)
+
+    def _process_spreadsheet(self, content: bytes, filename: str) -> str:
+        text_blocks = []
+        try:
+            if filename.endswith(".csv"):
+                df = pd.read_csv(io.BytesIO(content))
+                text_blocks.append(f"--- CSV Data ---\n{df.to_markdown(index=False)}")
+            else:
+                excel_file = pd.ExcelFile(io.BytesIO(content))
+                for sheet_name in excel_file.sheet_names:
+                    df = pd.read_excel(excel_file, sheet_name=sheet_name)
+                    text_blocks.append(f"--- Sheet: {sheet_name} ---\n{df.to_markdown(index=False)}")
+        except Exception as e:
+            return f"Error parsing spreadsheet: {e}"
+            
+        return "\n\n".join(text_blocks)
+
+file_service = FileService()

app/services/intent.py

@@ -0,0 +1,63 @@
+from typing import Literal
+from pydantic import BaseModel
+from app.services.llm import LLMService
+
+class IntentResponse(BaseModel):
+    category: Literal["SEARCH_REQUIRED", "CHAT_ONLY", "CODING_TASK", "DATA_ANALYSIS"]
+    reasoning: str
+    risk_level: Literal["LOW", "MEDIUM", "HIGH"]
+
+class IntentService:
+    def __init__(self, llm_service: LLMService):
+        self.llm = llm_service
+        self.system_prompt = """
+        You are the 'Intent Analyzer' for an AI Operating System.
+        Your job is to route the user's request to the correct module.
+        
+        Analyze the USER QUERY and return a JSON object.
+
+        CATEGORIES:
+        - SEARCH_REQUIRED: Query asks for current events, news, specific facts not in general knowledge, or research. (e.g., "Stock price of Apple", "Latest AI papers")
+        - CHAT_ONLY: General greetings, philosophical questions, summaries of previous context, or logic puzzles. (e.g., "Hi", "Explain Stoicism")
+        - CODING_TASK: Requests to write, debug, or explain code.
+        - DATA_ANALYSIS: Requests involving CSVs, charts, or math aggregations.
+
+        RISK LEVELS:
+        - HIGH: Asking for dangerous/illegal content, PII, or financial advice.
+        - MEDIUM: Ambiguous queries or potential controversies.
+        - LOW: Safe, standard queries.
+
+        Output format: {"category": "...", "reasoning": "...", "risk_level": "..."}
+        """
+
+    async def analyze(self, query: str) -> IntentResponse:
+        # For efficiency, we can use a faster/smaller model or just the main one with strictly low temp
+        prompt = f"{self.system_prompt}\n\nUSER QUERY: {query}"
+        
+        # We'll rely on the main LLM to parse this for now.
+        # ideally this uses a 'router' model (cheap/fast)
+        response_text = await self.llm._generate(
+            messages=[{"role": "user", "content": prompt}],
+            temperature=0.0
+        )
+        
+        # Simple parsing logic (robustness would require structured output mode or regex)
+        # Assuming LLM behaves well with JSON instructions.
+        import json
+        import re
+
+        try:
+            # Clean markdown code blocks if present
+            clean_text = re.sub(r"```json|```", "", response_text).strip()
+            data = json.loads(clean_text)
+            return IntentResponse(**data)
+        except Exception as e:
+            # Fallback to Safe Default
+            print(f"Intent Parsing Failed: {e}. Defaulting to SEARCH.")
+            return IntentResponse(
+                category="SEARCH_REQUIRED", 
+                reasoning="Parsing error, defaulting to search.",
+                risk_level="low"
+            )
+
+intent_service = None # initialized in main/deps

app/services/llm.py

@@ -0,0 +1,240 @@
+from openai import AsyncOpenAI
+from app.core.config import settings
+from app.core import prompts
+import asyncio
+
+# Optional Hugging Face Client
+try:
+    from huggingface_hub import AsyncInferenceClient
+except ImportError:
+    AsyncInferenceClient = None
+
+# Optional Groq Client
+try:
+    from groq import AsyncGroq
+except ImportError:
+    AsyncGroq = None
+
+# Optional Gemini Client
+try:
+    import google.generativeai as genai
+except ImportError:
+    genai = None
+
+class LLMService:
+    def __init__(self):
+        self.openai_key = settings.OPENAI_API_KEY
+        self.hf_key = settings.HUGGINGFACE_API_KEY
+        self.groq_key = settings.GROQ_API_KEY
+        self.gemini_key = settings.GEMINI_API_KEY
+        
+        with open("debug.log", "a") as f:
+            f.write(f"\n--- INIT ---\n")
+            f.write(f"OpenAI Key: {bool(self.openai_key)}\n")
+            f.write(f"HF Key: {self.hf_key[:5] if self.hf_key else 'None'}\n")
+            f.write(f"Groq Key: {self.groq_key[:5] if self.groq_key else 'None'}\n")
+            f.write(f"Gemini Key: {self.gemini_key[:5] if self.gemini_key else 'None'}\n")
+
+        self.provider = "mock"
+        self.client = None
+        self.model = "gpt-4-turbo-preview"
+
+        # 1. Prefer Groq (User Requested - Fastest)
+        if self.groq_key and "gsk_" in self.groq_key and AsyncGroq:
+            self.provider = "groq"
+            self.client = AsyncGroq(api_key=self.groq_key)
+            self.model = "llama-3.3-70b-versatile"
+            print(f"LLM Provider: Groq ({self.model})")
+            with open("debug.log", "a") as f: f.write(f"Provider Selected: Groq ({self.model})\n")
+
+        # 2. Fallback to Gemini
+        elif self.gemini_key and genai:
+            self.provider = "gemini"
+            genai.configure(api_key=self.gemini_key)
+            
+            # Dynamically find a valid model
+            valid_model = "gemini-1.5-flash"
+            try:
+                available_models = [m.name for m in genai.list_models() if 'generateContent' in m.supported_generation_methods]
+                print(f"Available Gemini Models: {available_models}")
+                
+                # Rigid preference list
+                preferences = [
+                    "models/gemini-1.5-flash-latest",
+                    "models/gemini-1.5-flash-001",
+                    "models/gemini-2.0-flash-exp", 
+                    "models/gemini-1.5-pro-latest",
+                ]
+                
+                for pref in preferences:
+                    if pref in available_models:
+                        valid_model = pref
+                        break
+                    # specific check for partial match if exact not found?
+                    # sticking to exact or known good partials to avoid experimental/paid models
+                
+            except Exception as e:
+                print(f"Error listing Gemini models: {e}")
+
+            # self.client = genai.GenerativeModel(valid_model) # Do not bind static client
+            self.model = "auto-retry-gemini"
+            print(f"LLM Provider: Gemini (Auto-Retry Mode)")
+            with open("debug.log", "a") as f: f.write(f"Provider Selected: Gemini (Auto-Retry)\n")
+
+        # 3. Fallback logic removed/moved up
+        elif False: 
+            pass
+            # print(f"LLM Provider: Groq ({self.model})")
+
+        # 2. Fallback to Hugging Face
+        elif self.hf_key and "hf_" in self.hf_key and AsyncInferenceClient:
+            self.provider = "huggingface"
+            self.client = AsyncInferenceClient(token=self.hf_key)
+            self.model = "HuggingFaceH4/zephyr-7b-beta"
+            print(f"LLM Provider: Hugging Face ({self.model})")
+
+        # 3. Fallback to OpenAI
+        elif self.openai_key and "sk-" in self.openai_key:
+            self.provider = "openai"
+            self.client = AsyncOpenAI(api_key=self.openai_key)
+            print("LLM Provider: OpenAI")
+
+    async def _generate(self, messages, temperature=0.7):
+        """Helper to unify generation calls across providers"""
+        if self.provider == "openai":
+            response = await self.client.chat.completions.create(
+                model=self.model,
+                messages=messages,
+                temperature=temperature
+            )
+            return response.choices[0].message.content
+
+        elif self.provider == "gemini":
+            # Dynamic Retry for Gemini
+            fallback_models = [
+                "models/gemini-1.5-flash",
+                "models/gemini-1.5-flash-latest",
+                "models/gemini-1.5-flash-001",
+                "models/gemini-2.0-flash-exp",
+                "models/gemini-2.0-flash",
+                "models/gemini-1.5-pro",
+                "models/gemini-1.5-pro-latest",
+            ]
+
+            full_prompt = ""
+            for m in messages:
+                role_prefix = "System" if m['role'] == "system" else "User"
+                full_prompt += f"{role_prefix}: {m['content']}\n\n"
+
+            last_error = None
+            for model_name in fallback_models:
+                try:
+                    # Create client on the fly for each model
+                    client = genai.GenerativeModel(model_name)
+                    response = await client.generate_content_async(
+                        full_prompt,
+                        generation_config=genai.types.GenerationConfig(
+                            temperature=temperature
+                        )
+                    )
+                    return response.text
+                except Exception as e:
+                    print(f"Gemini Model {model_name} failed: {e}")
+                    last_error = e
+            
+            # If all fail, raise the last error to be caught by the caller
+            raise last_error
+
+        elif self.provider == "groq":
+            try:
+                response = await self.client.chat.completions.create(
+                    model=self.model,
+                    messages=messages,
+                    temperature=temperature
+                )
+                return response.choices[0].message.content
+            except Exception as e:
+                print(f"Groq Error: {e}")
+                with open("debug.log", "a") as f: f.write(f"Groq Error: {e}\n")
+                raise e
+            
+        elif self.provider == "huggingface":
+            # HF AsyncInferenceClient also supports chat_completion style
+            response = await self.client.chat_completion(
+                model=self.model,
+                messages=messages,
+                temperature=temperature,
+                max_tokens=1000
+            )
+            return response.choices[0].message.content
+            
+        return None
+
+    async def get_search_queries(self, user_query: str) -> str:
+        if self.provider == "mock":
+             return f"Search Query: {user_query}"
+
+        try:
+            messages = [
+                {"role": "system", "content": prompts.SYSTEM_PROMPT},
+                {"role": "system", "content": prompts.MASTER_PROMPT_SEARCH},
+                {"role": "user", "content": user_query}
+            ]
+            content = await self._generate(messages, temperature=0.2)
+            return content if content else user_query
+        except Exception as e:
+            with open("debug.log", "a") as f: f.write(f"LLM Error (Search): {e}\n")
+            return user_query
+
+    async def synthesize_answer(self, user_query: str, sources_text: str) -> str:
+        if self.provider == "mock":
+            await asyncio.sleep(1)
+            return self._mock_synthesize(user_query, sources_text)
+
+        try:
+            messages = [
+                {"role": "system", "content": prompts.SYSTEM_PROMPT},
+                {"role": "system", "content": prompts.MASTER_PROMPT_ANSWER},
+                {"role": "user", "content": f"User Question: {user_query}\n\nExisting Sources:\n{sources_text}"}
+            ]
+            content = await self._generate(messages, temperature=0.1)
+            return content if content else self._mock_synthesize(user_query, sources_text)
+        except Exception as e:
+            with open("debug.log", "a") as f: f.write(f"LLM Error (Synthesis): {e}\n")
+            return self._mock_synthesize(user_query, sources_text)
+
+    async def verify_confidence(self, answer: str, sources_text: str) -> str:
+        if self.provider == "mock":
+            return "High confidence. (Mock Verification)"
+
+        try:
+            messages = [
+                {"role": "system", "content": prompts.MASTER_PROMPT_CONFIDENCE},
+                {"role": "user", "content": f"Answer to evaluate:\n{answer}\n\nBased on sources:\n{sources_text}"}
+            ]
+            content = await self._generate(messages, temperature=0.1)
+            return content if content else "Medium confidence"
+        except Exception as e:
+            print(f"LLM Error (Confidence): {e}")
+            return "Medium confidence"
+
+    def _mock_synthesize(self, query: str, sources_text: str = "") -> str:
+        source_preview = "No sources found."
+        if sources_text:
+             lines = sources_text.split('\n')
+             titles = [line.replace("Title: ", "") for line in lines if line.startswith("Title: ")]
+             source_preview = f"Found {len(titles)} Sources including: " + ", ".join(titles[:3])
+
+        return f"""
+**Note: Running in Partial Mode (Tavily Connected, No LLM Key).**
+
+The system successfully searched using your **Tavily Key** and found real data:
+_{source_preview}_
+
+However, I cannot generate the answer. 
+Current Provider: **{self.provider}**
+Check debug.log for specific error (e.g., 404 Model Not Found or Auth Error).
+"""
+
+llm_service = LLMService()
+

app/services/search.py

@@ -0,0 +1,55 @@
+import os
+from typing import List, Dict, Any
+from app.core.config import settings
+
+# Placeholder for Tavily or similar search client
+try:
+    from tavily import TavilyClient
+except ImportError:
+    TavilyClient = None
+
+class SearchService:
+    def __init__(self):
+        self.api_key = settings.TAVILY_API_KEY
+        self.client = None
+        if self.api_key and TavilyClient:
+            self.client = TavilyClient(api_key=self.api_key)
+
+    async def search(self, query: str, search_depth: str = "advanced") -> List[Dict[str, Any]]:
+        """
+        Executes a search using the Master Prompt strategies (conceptually).
+        In reality, we send the query to a search API.
+        """
+        print(f"Executing search for: {query}")
+        
+        if not self.client:
+            # Mock response for testing without API Key
+            return self._mock_search_results(query)
+
+        try:
+            # Tavily context search is optimized for RAG
+            response = self.client.search(query=query, search_depth=search_depth, include_domains=[])
+            return response.get("results", [])
+        except Exception as e:
+            print(f"Search failed: {e}")
+            return []
+
+    def _mock_search_results(self, query: str) -> List[Dict[str, Any]]:
+        return [
+            {
+                "title": f"Mock Source about {query}",
+                "url": "https://example.com/source1",
+                "content": f"This is a verified source containing specific data about {query}. It confirms the primary query intent.",
+                "score": 0.95,
+                "published_date": "2024-01-01"
+            },
+            {
+                "title": f"Another Relevant Source for {query}",
+                "url": "https://gov.data/report",
+                "content": f"Official government data suggests that {query} has seen a 20% increase in importance.",
+                "score": 0.88,
+                "published_date": "2023-12-15"
+            }
+        ]
+
+search_service = SearchService()

app/services/vector.py

@@ -0,0 +1,76 @@
+import numpy as np
+try:
+    import faiss
+    from sentence_transformers import SentenceTransformer
+except Exception as e:
+    print(f"VectorService: Failed to import dependencies: {e}")
+    faiss = None
+    SentenceTransformer = None
+
+class VectorService:
+    def __init__(self):
+        self.model = None
+        self.index = None
+        self.chunks = []
+        
+        if SentenceTransformer:
+            # Load model once. This might be slow on startup.
+            print("Loading generic embedding model (all-MiniLM-L6-v2)...")
+            try:
+                self.model = SentenceTransformer("all-MiniLM-L6-v2")
+                print("Embedding model loaded successfully.")
+            except Exception as e:
+                print(f"Failed to load embedding model: {e}")
+    
+    def create_index_from_results(self, results: list):
+        """
+        Takes a list of search result dicts, creates embeddings, and builds a FAISS index.
+        """
+        if not self.model or not faiss:
+            print("VectorService: Dependencies missing or model not loaded.")
+            return
+
+        self.chunks = []
+        texts_to_embed = []
+        
+        for res in results:
+            # Combine Title and Content for a rich embedding context
+            text = f"Title: {res.get('title', '')}\nContent: {res.get('content', '')}"
+            self.chunks.append(res) # Keep reference to original object
+            texts_to_embed.append(text)
+            
+        if not texts_to_embed:
+            return
+
+        try:
+            embeddings = self.model.encode(texts_to_embed)
+            dimension = embeddings.shape[1]
+            
+            self.index = faiss.IndexFlatL2(dimension)
+            self.index.add(np.array(embeddings))
+            print(f"VectorService: Created FAISS index with {self.index.ntotal} vectors")
+        except Exception as e:
+            print(f"VectorService Error during indexing: {e}")
+
+    def search_similar(self, query: str, k: int = 3):
+        """
+        Searches the FAISS index for the most relevant chunks to the query.
+        """
+        if not self.index or not self.model:
+            return []
+            
+        try:
+            query_emb = self.model.encode([query])
+            distances, indices = self.index.search(query_emb, k)
+            
+            top_results = []
+            for idx in indices[0]:
+                if idx < len(self.chunks) and idx >= 0:
+                    top_results.append(self.chunks[idx])
+                    
+            return top_results
+        except Exception as e:
+            print(f"VectorService Error during search: {e}")
+            return []
+
+vector_service = VectorService()

backend/.env

@@ -0,0 +1,17 @@
+# Trust-First Copilot Configuration
+
+# OpenAI API Key (Optional if using Hugging Face)
+OPENAI_API_KEY=
+
+# Tavily API Key (Required for Search)
+TAVILY_API_KEY=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
+
+# Hugging Face API Key (Free Brain Power!)
+HUGGINGFACE_API_KEY=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
+
+# Groq API Key (Fastest Brain)
+GROQ_API_KEY=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
+PROJECT_NAME="Advance Copilot"
+
+# Gemini API Key (Google)
+GEMINI_API_KEY=AIzaSyBb27_DT8Py1m2KrctHNZuIcuyu75xWOfo

backend/debug.log

@@ -0,0 +1,823 @@
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+LLM Error (Synthesis): Error code: 404 - {'error': {'message': 'The model `gpt-4-turbo-preview` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
+LLM Error (Synthesis): Error code: 404 - {'error': {'message': 'The model `gpt-4-turbo-preview` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
+LLM Error (Synthesis): Error code: 404 - {'error': {'message': 'The model `gpt-4-turbo-preview` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
+LLM Error (Synthesis): Error code: 404 - {'error': {'message': 'The model `gpt-4-turbo-preview` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
+LLM Error (Synthesis): Error code: 404 - {'error': {'message': 'The model `gpt-4-turbo-preview` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+LLM Error (Synthesis): (Request ID: Root=1-696658c4-788fc3276bd8d2ba39479c5d;254be642-c222-4663-84d5-f8ba42f2ebbe)
+
+403 Forbidden: This authentication method does not have sufficient permissions to call Inference Providers on behalf of user Abhi18singh.
+Cannot access content at: https://router.huggingface.co/v1/chat/completions.
+Make sure your token has the correct permissions.
+LLM Error (Synthesis): (Request ID: Root=1-696658ef-5a3811376f5b88c367c230fc;3c7bda51-0042-48c4-9c9d-9cd4f4e4d6d2)
+
+403 Forbidden: This authentication method does not have sufficient permissions to call Inference Providers on behalf of user Abhi18singh.
+Cannot access content at: https://router.huggingface.co/v1/chat/completions.
+Make sure your token has the correct permissions.
+LLM Error (Synthesis): (Request ID: Root=1-69665998-7a2fa3962140a6b8222dd600;b74da982-0738-479c-9651-854274a658d2)
+
+403 Forbidden: This authentication method does not have sufficient permissions to call Inference Providers on behalf of user Abhi18singh.
+Cannot access content at: https://router.huggingface.co/v1/chat/completions.
+Make sure your token has the correct permissions.
+LLM Error (Synthesis): (Request ID: Root=1-69665a20-1573ab002b04c9556e6c0c59;fa51d8f9-7c0a-4211-aacd-105bec200581)
+
+403 Forbidden: This authentication method does not have sufficient permissions to call Inference Providers on behalf of user Abhi18singh.
+Cannot access content at: https://router.huggingface.co/v1/chat/completions.
+Make sure your token has the correct permissions.
+LLM Error (Synthesis): (Request ID: Root=1-69665a5e-0646812e1eec86ba4943610e;25d67883-df55-4879-869e-639ffc2a2eb8)
+
+403 Forbidden: This authentication method does not have sufficient permissions to call Inference Providers on behalf of user Abhi18singh.
+Cannot access content at: https://router.huggingface.co/v1/chat/completions.
+Make sure your token has the correct permissions.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+AsyncInferenceClient: <class 'huggingface_hub.inference._generated._async_client.AsyncInferenceClient'>
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+LLM Error (Synthesis): Error code: 400 - {'error': {'message': 'The model `llama3-70b-8192` has been decommissioned and is no longer supported. Please refer to https://console.groq.com/docs/deprecations for a recommendation on which model to use instead.', 'type': 'invalid_request_error', 'code': 'model_decommissioned'}}
+LLM Error (Synthesis): Error code: 400 - {'error': {'message': 'The model `llama3-70b-8192` has been decommissioned and is no longer supported. Please refer to https://console.groq.com/docs/deprecations for a recommendation on which model to use instead.', 'type': 'invalid_request_error', 'code': 'model_decommissioned'}}
+LLM Error (Synthesis): Error code: 400 - {'error': {'message': 'The model `llama3-70b-8192` has been decommissioned and is no longer supported. Please refer to https://console.groq.com/docs/deprecations for a recommendation on which model to use instead.', 'type': 'invalid_request_error', 'code': 'model_decommissioned'}}
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): Error code: 400 - {'error': {'message': 'The model `llama3-70b-8192` has been decommissioned and is no longer supported. Please refer to https://console.groq.com/docs/deprecations for a recommendation on which model to use instead.', 'type': 'invalid_request_error', 'code': 'model_decommissioned'}}
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.5-pro
+Please retry in 24.311506491s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerMinute-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerMinutePerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerDay-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+, retry_delay {
+  seconds: 24
+}
+]
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.5-pro
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.5-pro
+Please retry in 42.395475103s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerDay-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerMinutePerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerMinute-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-pro"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+, retry_delay {
+  seconds: 42
+}
+]
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.0-flash
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash
+Please retry in 50.635719043s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerMinute-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerMinutePerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+, retry_delay {
+  seconds: 50
+}
+]
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.0-flash
+Please retry in 6.198648352s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerMinutePerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerMinute-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+, retry_delay {
+  seconds: 6
+}
+]
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 20, model: gemini-2.5-flash
+Please retry in 9.804671677s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+  quota_value: 20
+}
+, retry_delay {
+  seconds: 9
+}
+]
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 20, model: gemini-2.5-flash
+Please retry in 29.659097469s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+  quota_value: 20
+}
+, retry_delay {
+  seconds: 29
+}
+]
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 20, model: gemini-2.5-flash
+Please retry in 3.690783496s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.5-flash"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+  quota_value: 20
+}
+, retry_delay {
+  seconds: 3
+}
+]
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 404 models/gemini-1.5-flash is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+LLM Error (Synthesis): 429 You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. 
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash-exp
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_requests, limit: 0, model: gemini-2.0-flash-exp
+* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 0, model: gemini-2.0-flash-exp
+Please retry in 36.195734558s. [links {
+  description: "Learn more about Gemini API quotas"
+  url: "https://ai.google.dev/gemini-api/docs/rate-limits"
+}
+, violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerDayPerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash-exp"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_requests"
+  quota_id: "GenerateRequestsPerMinutePerProjectPerModel-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash-exp"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+violations {
+  quota_metric: "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count"
+  quota_id: "GenerateContentInputTokensPerModelPerMinute-FreeTier"
+  quota_dimensions {
+    key: "model"
+    value: "gemini-2.0-flash-exp"
+  }
+  quota_dimensions {
+    key: "location"
+    value: "global"
+  }
+}
+, retry_delay {
+  seconds: 36
+}
+]
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)
+Provider Selected: Groq (llama-3.3-70b-versatile)
+Provider Selected: Groq (llama-3.3-70b-versatile)
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)
+
+--- INIT ---
+OpenAI Key: True
+HF Key: hf_Js
+Groq Key: gsk_f
+Gemini Key: AIzaS
+Provider Selected: Groq (llama-3.3-70b-versatile)

backend/debug_env.py

@@ -0,0 +1,47 @@
+import os
+from dotenv import load_dotenv
+load_dotenv()
+
+print("--- ENV DEBUG ---")
+hf_key = os.getenv("HUGGINGFACE_API_KEY")
+print(f"HF Key present: {bool(hf_key)}")
+print(f"HF Key start: {hf_key[:4] if hf_key else 'None'}")
+
+print("\n--- LIBRARY DEBUG ---")
+try:
+    import huggingface_hub
+    print(f"huggingface_hub version: {huggingface_hub.__version__}")
+except ImportError:
+    print("huggingface_hub NOT installed")
+
+try:
+    from huggingface_hub import AsyncInferenceClient
+    print("AsyncInferenceClient imported successfully")
+except ImportError as e:
+    print(f"AsyncInferenceClient import FAILED: {e}")
+
+try:
+    from huggingface_hub import InferenceClient
+    print("InferenceClient imported successfully")
+except ImportError as e:
+    print(f"InferenceClient import FAILED: {e}")
+
+print("\n--- VECTOR DB DEBUG ---")
+try:
+    import faiss
+    print(f"FAISS imported successfully")
+except ImportError:
+    print("FAISS NOT installed")
+
+try:
+    from sentence_transformers import SentenceTransformer
+    print("SentenceTransformer imported successfully")
+except ImportError:
+    print("SentenceTransformer NOT installed")
+
+print("\n--- APP CONFIG DEBUG ---")
+try:
+    from app.core.config import settings
+    print(f"Settings HF Key: {bool(settings.HUGGINGFACE_API_KEY)}")
+except Exception as e:
+    print(f"Could not load settings: {e}")

backend/list_available_models.py

@@ -0,0 +1,21 @@
+import google.generativeai as genai
+import os
+from dotenv import load_dotenv
+
+load_dotenv()
+api_key = os.getenv("GEMINI_API_KEY")
+
+if not api_key:
+    print("No API Key found")
+    exit()
+
+print(f"Checking models for key: {api_key[:10]}...")
+genai.configure(api_key=api_key)
+
+try:
+    print("List of available models:")
+    for m in genai.list_models():
+        if 'generateContent' in m.supported_generation_methods:
+            print(f"- {m.name}")
+except Exception as e:
+    print(f"Error listing models: {e}")

backend/list_models.py

@@ -0,0 +1,12 @@
+import google.generativeai as genai
+from app.core.config import settings
+
+genai.configure(api_key=settings.GEMINI_API_KEY)
+
+print("Listing models...")
+try:
+    for m in genai.list_models():
+        if 'generateContent' in m.supported_generation_methods:
+            print(m.name)
+except Exception as e:
+    print(f"Error listing models: {e}")

backend/rag_summarizer.py

@@ -0,0 +1,81 @@
+from dotenv import load_dotenv
+import os
+import google.generativeai as genai
+from tavily import TavilyClient
+
+# =====================
+# LOAD ENV
+# =====================
+load_dotenv()
+
+GEMINI_KEY = os.getenv("GEMINI_API_KEY")
+TAVILY_KEY = os.getenv("TAVILY_API_KEY")
+
+print(f"Gemini Key: {GEMINI_KEY[:10]}...") 
+
+if not GEMINI_KEY:
+    print("❌ Gemini API key not loaded")
+    exit(1)
+
+# =====================
+# INIT SERVICES
+# =====================
+genai.configure(api_key=GEMINI_KEY)
+model = genai.GenerativeModel("gemini-1.5-flash")
+
+# Mock Tavily if key missing (for testing Gemini specifically)
+if TAVILY_KEY:
+    tavily = TavilyClient(api_key=TAVILY_KEY)
+else:
+    tavily = None
+    print("Warning: Tavily Key missing, using mock data")
+
+# =====================
+# USER QUERY
+# =====================
+query = "AI in healthcare current and future applications"
+
+# =====================
+# SEARCH (TAVILY)
+# =====================
+if tavily:
+    search_results = tavily.search(
+        query=query,
+        max_results=5,
+        include_raw_content=True
+    )
+    context = "\n\n".join(
+        [res["content"] for res in search_results["results"]]
+    )
+else:
+    context = "Artificial Intelligence involves using computers to simulate human intelligence. AI is used in healthcare for diagnosis and treatment planning."
+
+# =====================
+# PROMPT
+# =====================
+PROMPT = f"""
+You are an expert research assistant.
+
+Task:
+- Generate a concise, accurate, and well-structured answer
+- Use ONLY the information from the sources below
+- If unsure, say "Information not found in sources"
+
+Sources:
+{context}
+
+Instructions:
+- Summarize in 6–8 bullet points
+- Use simple professional language
+- Focus on real-world applications and future impact
+"""
+
+# =====================
+# GENERATE ANSWER
+# =====================
+try:
+    response = model.generate_content(PROMPT)
+    print("\n✅ FINAL ANSWER:\n")
+    print(response.text)
+except Exception as e:
+    print(f"\n❌ ERROR: {e}")

debug.log

@@ -0,0 +1,18 @@
+
+--- INIT ---
+OpenAI Key: True
+HF Key: None
+Groq Key: None
+Gemini Key: None
+
+--- INIT ---
+OpenAI Key: True
+HF Key: None
+Groq Key: None
+Gemini Key: None
+
+--- INIT ---
+OpenAI Key: True
+HF Key: None
+Groq Key: None
+Gemini Key: None

frontend/.gitignore

@@ -0,0 +1,41 @@
+# See https://help.github.com/articles/ignoring-files/ for more about ignoring files.
+
+# dependencies
+/node_modules
+/.pnp
+.pnp.*
+.yarn/*
+!.yarn/patches
+!.yarn/plugins
+!.yarn/releases
+!.yarn/versions
+
+# testing
+/coverage
+
+# next.js
+/.next/
+/out/
+
+# production
+/build
+
+# misc
+.DS_Store
+*.pem
+
+# debug
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+.pnpm-debug.log*
+
+# env files (can opt-in for committing if needed)
+.env*
+
+# vercel
+.vercel
+
+# typescript
+*.tsbuildinfo
+next-env.d.ts

frontend/README.md

@@ -0,0 +1,36 @@
+This is a [Next.js](https://nextjs.org) project bootstrapped with [`create-next-app`](https://nextjs.org/docs/app/api-reference/cli/create-next-app).
+
+## Getting Started
+
+First, run the development server:
+
+```bash
+npm run dev
+# or
+yarn dev
+# or
+pnpm dev
+# or
+bun dev
+```
+
+Open [http://localhost:3000](http://localhost:3000) with your browser to see the result.
+
+You can start editing the page by modifying `app/page.tsx`. The page auto-updates as you edit the file.
+
+This project uses [`next/font`](https://nextjs.org/docs/app/building-your-application/optimizing/fonts) to automatically optimize and load [Geist](https://vercel.com/font), a new font family for Vercel.
+
+## Learn More
+
+To learn more about Next.js, take a look at the following resources:
+
+- [Next.js Documentation](https://nextjs.org/docs) - learn about Next.js features and API.
+- [Learn Next.js](https://nextjs.org/learn) - an interactive Next.js tutorial.
+
+You can check out [the Next.js GitHub repository](https://github.com/vercel/next.js) - your feedback and contributions are welcome!
+
+## Deploy on Vercel
+
+The easiest way to deploy your Next.js app is to use the [Vercel Platform](https://vercel.com/new?utm_medium=default-template&filter=next.js&utm_source=create-next-app&utm_campaign=create-next-app-readme) from the creators of Next.js.
+
+Check out our [Next.js deployment documentation](https://nextjs.org/docs/app/building-your-application/deploying) for more details.

frontend/app/globals.css

@@ -0,0 +1,100 @@
+@import "tailwindcss";
+
+
+@theme {
+  --color-background: var(--background);
+  --color-foreground: var(--foreground);
+  --color-card: var(--card);
+  --color-card-foreground: var(--card-foreground);
+  --color-popover: var(--popover);
+  --color-popover-foreground: var(--popover-foreground);
+  --color-primary: var(--primary);
+  --color-primary-foreground: var(--primary-foreground);
+  --color-secondary: var(--secondary);
+  --color-secondary-foreground: var(--secondary-foreground);
+  --color-muted: var(--muted);
+  --color-muted-foreground: var(--muted-foreground);
+  --color-accent: var(--accent);
+  --color-accent-foreground: var(--accent-foreground);
+  --color-destructive: var(--destructive);
+  --color-destructive-foreground: var(--destructive-foreground);
+  --color-border: var(--border);
+  --color-input: var(--input);
+  --color-ring: var(--ring);
+
+  --radius-lg: var(--radius);
+  --radius-md: calc(var(--radius) - 2px);
+  --radius-sm: calc(var(--radius) - 4px);
+}
+
+:root {
+  --background: 0 0% 100%;
+  --foreground: 222.2 84% 4.9%;
+
+  --card: 0 0% 100%;
+  --card-foreground: 222.2 84% 4.9%;
+
+  --popover: 0 0% 100%;
+  --popover-foreground: 222.2 84% 4.9%;
+
+  --primary: 222.2 47.4% 11.2%;
+  --primary-foreground: 210 40% 98%;
+
+  --secondary: 210 40% 96.1%;
+  --secondary-foreground: 222.2 47.4% 11.2%;
+
+  --muted: 210 40% 96.1%;
+  --muted-foreground: 215.4 16.3% 46.9%;
+
+  --accent: 210 40% 96.1%;
+  --accent-foreground: 222.2 47.4% 11.2%;
+
+  --destructive: 0 84.2% 60.2%;
+  --destructive-foreground: 210 40% 98%;
+
+  --border: 214.3 31.8% 91.4%;
+  --input: 214.3 31.8% 91.4%;
+  --ring: 222.2 84% 4.9%;
+
+  --radius: 0.5rem;
+}
+
+.dark {
+  --background: 222.2 84% 4.9%;
+  --foreground: 210 40% 98%;
+
+  --card: 222.2 84% 4.9%;
+  --card-foreground: 210 40% 98%;
+
+  --popover: 222.2 84% 4.9%;
+  --popover-foreground: 210 40% 98%;
+
+  --primary: 210 40% 98%;
+  --primary-foreground: 222.2 47.4% 11.2%;
+
+  --secondary: 217.2 32.6% 17.5%;
+  --secondary-foreground: 210 40% 98%;
+
+  --muted: 217.2 32.6% 17.5%;
+  --muted-foreground: 215 20.2% 65.1%;
+
+  --accent: 217.2 32.6% 17.5%;
+  --accent-foreground: 210 40% 98%;
+
+  --destructive: 0 62.8% 30.6%;
+  --destructive-foreground: 210 40% 98%;
+
+  --border: 217.2 32.6% 17.5%;
+  --input: 217.2 32.6% 17.5%;
+  --ring: 212.7 26.8% 83.9%;
+}
+
+@layer base {
+  * {
+    @apply border-border;
+  }
+
+  body {
+    @apply bg-background text-foreground;
+  }
+}

frontend/app/layout.tsx

@@ -0,0 +1,35 @@
+import type { Metadata } from "next";
+import { Geist, Geist_Mono } from "next/font/google";
+import "./globals.css";
+
+const geistSans = Geist({
+  variable: "--font-geist-sans",
+  subsets: ["latin"],
+});
+
+const geistMono = Geist_Mono({
+  variable: "--font-geist-mono",
+  subsets: ["latin"],
+});
+
+export const metadata: Metadata = {
+  title: "Create Next App",
+  description: "Generated by create next app",
+};
+
+export default function RootLayout({
+  children,
+}: Readonly<{
+  children: React.ReactNode;
+}>) {
+  return (
+    <html lang="en" suppressHydrationWarning>
+      <body
+        className={`${geistSans.variable} ${geistMono.variable} antialiased`}
+        suppressHydrationWarning
+      >
+        {children}
+      </body>
+    </html>
+  );
+}

frontend/app/page.tsx

@@ -0,0 +1,5 @@
+import ChatLayout from "../components/Chat/ChatLayout";
+
+export default function Home() {
+  return <ChatLayout />;
+}

frontend/components/Chat/ChatInput.tsx

@@ -0,0 +1,87 @@
+"use client";
+
+import { SendHorizontal, Paperclip } from "lucide-react";
+import { useRef, useEffect } from "react";
+
+interface ChatInputProps {
+    value: string;
+    onChange: (val: string) => void;
+    onSubmit: () => void;
+    isLoading: boolean;
+    onFileSelect?: (file: File) => void;
+    attachedFile?: File | null;
+}
+
+export default function ChatInput({ value, onChange, onSubmit, isLoading, onFileSelect, attachedFile }: ChatInputProps) {
+    const textareaRef = useRef<HTMLTextAreaElement>(null);
+    const fileInputRef = useRef<HTMLInputElement>(null);
+
+    // Auto-resize textarea
+    useEffect(() => {
+        if (textareaRef.current) {
+            textareaRef.current.style.height = "auto";
+            textareaRef.current.style.height = `${Math.min(textareaRef.current.scrollHeight, 200)}px`;
+        }
+    }, [value]);
+
+    const handleKeyDown = (e: React.KeyboardEvent) => {
+        if (e.key === "Enter" && !e.shiftKey) {
+            e.preventDefault();
+            onSubmit();
+        }
+    };
+
+    return (
+        <div className="w-full max-w-3xl mx-auto px-4 pb-6">
+            <div className="relative flex items-end gap-2 bg-white border border-gray-200 rounded-xl p-3 shadow-lg focus-within:ring-2 focus-within:ring-indigo-500/50 focus-within:border-indigo-500 transition-all">
+                <input
+                    type="file"
+                    ref={fileInputRef}
+                    className="hidden"
+                    onChange={(e) => {
+                        if (e.target.files?.[0] && onFileSelect) onFileSelect(e.target.files[0]);
+                    }}
+                />
+                <button
+                    onClick={() => fileInputRef.current?.click()}
+                    className={`p-2 transition-colors ${attachedFile ? "text-indigo-600 bg-indigo-50 rounded" : "text-gray-400 hover:text-gray-600"}`}
+                    title="Attach file (PDF, Excel, Txt)"
+                >
+                    <Paperclip size={20} />
+                </button>
+
+                <textarea
+                    ref={textareaRef}
+                    value={value}
+                    onChange={(e) => onChange(e.target.value)}
+                    onKeyDown={handleKeyDown}
+                    placeholder="Ask anything..."
+                    className="flex-1 bg-transparent border-0 focus:ring-0 text-gray-900 placeholder-gray-400 resize-none max-h-[200px] min-h-[24px] py-2"
+                    disabled={isLoading}
+                    rows={1}
+                />
+
+                <button
+                    onClick={onSubmit}
+                    disabled={isLoading || !value.trim()}
+                    className={`p-2 rounded-lg transition-all ${value.trim() && !isLoading
+                        ? "bg-indigo-600 text-white hover:bg-indigo-700"
+                        : "bg-gray-100 text-gray-300 cursor-not-allowed"
+                        }`}
+                >
+                    <SendHorizontal size={20} />
+                </button>
+            </div>
+            {attachedFile && (
+                <div className="mt-2 ml-1 text-xs text-indigo-600 flex items-center gap-1 font-medium animate-fade-in">
+                    <span>📄 Attached: {attachedFile.name} (will be analyzed)</span>
+                </div>
+            )}
+            <div className="text-center mt-2">
+                <p className="text-xs text-gray-400">
+                    Trust-First Copilot can make mistakes. Check important info.
+                </p>
+            </div>
+        </div>
+    );
+}

frontend/components/Chat/ChatLayout.tsx

@@ -0,0 +1,237 @@
+"use client";
+
+import { useState, useEffect } from "react";
+import Sidebar from "./Sidebar";
+import MessageBubble from "./MessageBubble";
+import ChatInput from "./ChatInput";
+import { Menu } from "lucide-react";
+
+interface Message {
+    role: "user" | "assistant";
+    content: string;
+    sources?: any[];
+    intent?: string;
+    thought_process?: string;
+}
+
+interface Session {
+    id: string;
+    title: string;
+    date: string;
+}
+
+export default function ChatLayout() {
+    const [isSidebarOpen, setIsSidebarOpen] = useState(true);
+    const [query, setQuery] = useState("");
+    const [attachedFile, setAttachedFile] = useState<File | null>(null);
+    const [messages, setMessages] = useState<Message[]>([]);
+    const [isLoading, setIsLoading] = useState(false);
+    const [sessions, setSessions] = useState<Session[]>([]);
+    const [currentSessionId, setCurrentSessionId] = useState<string | null>(null);
+
+    // Load history from local storage on mount
+    useEffect(() => {
+        const saved = localStorage.getItem("chat_sessions");
+        if (saved) {
+            setSessions(JSON.parse(saved));
+        }
+    }, []);
+
+    const handleNewChat = () => {
+        setMessages([]);
+        setCurrentSessionId(null);
+    };
+
+    const handleSearch = async () => {
+        if (!query.trim()) return;
+
+        // Add User Message
+        const userMsg: Message = { role: "user", content: query };
+        setMessages((prev) => [...prev, userMsg]);
+        setQuery("");
+        setIsLoading(true);
+
+        try {
+            let finalQuery = userMsg.content;
+
+            // Upload File if exists
+            if (attachedFile) {
+                const formData = new FormData();
+                formData.append("file", attachedFile);
+
+                // Optimistic UI update: Show file as uploaded
+                setMessages((prev) => [...prev, { role: "assistant", content: `📂 Analyzing ${attachedFile.name}...`, isLoading: true }]);
+
+                const uploadRes = await fetch(`${process.env.NEXT_PUBLIC_API_BASE || "http://localhost:8000"}/api/v1/upload`, {
+                    method: "POST",
+                    body: formData,
+                });
+                const uploadData = await uploadRes.json();
+
+                // Remove the loading message logic would be complex, simplfying:
+                // Append file context to query
+                finalQuery = `[Context from uploaded file ${attachedFile.name}]:\n${uploadData.content}\n\nUser Question: ${userMsg.content}`;
+                setAttachedFile(null); // Clear file
+            }
+
+            const response = await fetch(`${process.env.NEXT_PUBLIC_API_BASE || "http://localhost:8000"}/api/v1/query`, {
+                method: "POST",
+                headers: { "Content-Type": "application/json" },
+                body: JSON.stringify({ query: finalQuery }),
+            });
+
+            const data = await response.json();
+
+            // Add Assistant Message
+            const aiMsg: Message = {
+                role: "assistant",
+                content: data.answer,
+                sources: data.sources,
+                intent: data.intent,
+                thought_process: data.thought_process,
+            };
+            setMessages((prev) => [...prev, aiMsg]);
+
+            // Save Session (Simple Logic)
+            if (!currentSessionId) {
+                const newId = Date.now().toString();
+                const newSession = {
+                    id: newId,
+                    title: userMsg.content.slice(0, 30) + "...",
+                    date: new Date().toLocaleDateString(),
+                };
+                setSessions((prev) => [newSession, ...prev]);
+                setCurrentSessionId(newId);
+                localStorage.setItem("chat_sessions", JSON.stringify([newSession, ...sessions]));
+            }
+
+        } catch (error) {
+            console.error("Error:", error);
+            setMessages((prev) => [
+                ...prev,
+                { role: "assistant", content: "Sorry, something went wrong. Please try again." },
+            ]);
+        } finally {
+            setIsLoading(false);
+        }
+    };
+
+    const handleChallenge = async (msgIndex: number) => {
+        const targetMsg = messages[msgIndex];
+        if (!targetMsg || targetMsg.role !== "assistant") return;
+
+        // Find the preceding user message for context (simple heuristic: index - 1)
+        const userQuery = messages[msgIndex - 1]?.content || "Unknown context";
+        const sourcesText = targetMsg.sources
+            ? targetMsg.sources.map(s => `Title: ${s.title}\nContent: ${s.snippet}`).join("\n\n")
+            : "No sources.";
+
+        setIsLoading(true);
+
+        try {
+            const response = await fetch(`${process.env.NEXT_PUBLIC_API_BASE || "http://localhost:8000"}/api/v1/challenge`, {
+                method: "POST",
+                headers: { "Content-Type": "application/json" },
+                body: JSON.stringify({
+                    original_query: userQuery,
+                    original_answer: targetMsg.content,
+                    sources_text: sourcesText
+                }),
+            });
+
+            const data = await response.json();
+
+            // Add Critique Message
+            const critiqueMsg: Message = {
+                role: "assistant",
+                content: data.answer,
+                sources: [],
+                intent: "CRITIQUE",
+                thought_process: "Devil's Advocate Mode: Analyzing potential flaws in the previous answer."
+            };
+            setMessages((prev) => [...prev, critiqueMsg]);
+
+        } catch (error) {
+            console.error("Challenge Error:", error);
+        } finally {
+            setIsLoading(false);
+        }
+    };
+
+    return (
+        <div className="flex h-screen bg-white text-gray-900 font-sans overflow-hidden">
+            {/* Sidebar */}
+            <Sidebar
+                isOpen={isSidebarOpen}
+                sessions={sessions}
+                currentSessionId={currentSessionId}
+                onNewChat={handleNewChat}
+                onSelectSession={(id) => console.log("Select session:", id)} // Placeholder for loading specific session
+                onDeleteSession={(id, e) => {
+                    e.stopPropagation();
+                    const newSessions = sessions.filter(s => s.id !== id);
+                    setSessions(newSessions);
+                    localStorage.setItem("chat_sessions", JSON.stringify(newSessions));
+                    if (currentSessionId === id) handleNewChat();
+                }}
+            />
+
+            {/* Main Content */}
+            <div className="flex-1 flex flex-col h-full relative">
+                {/* Header / Mobile Toggle */}
+                <div className="absolute top-4 left-4 z-20">
+                    <button
+                        onClick={() => setIsSidebarOpen(!isSidebarOpen)}
+                        className="p-2 text-gray-400 hover:text-gray-900 rounded-lg hover:bg-gray-100 transition-colors"
+                    >
+                        <Menu size={20} />
+                    </button>
+                </div>
+
+                {/* Chat Area */}
+                <div className="flex-1 overflow-y-auto scroll-smooth">
+                    {messages.length === 0 ? (
+                        <div className="h-full flex flex-col items-center justify-center p-8 text-center opacity-50">
+                            <div className="w-16 h-16 bg-gray-100 rounded-2xl flex items-center justify-center mb-6">
+                                <span className="text-3xl">✨</span>
+                            </div>
+                            <h2 className="text-2xl font-semibold mb-2 text-gray-900">Trust-First Copilot</h2>
+                            <p className="max-w-md text-gray-500">
+                                Ask anything. I check reliable sources before answering.
+                            </p>
+                        </div>
+                    ) : (
+                        <div className="pb-32">
+                            {messages.map((msg, idx) => (
+                                <MessageBubble
+                                    key={idx}
+                                    role={msg.role}
+                                    content={msg.content}
+                                    sources={msg.sources}
+                                    intent={msg.intent}
+                                    thought_process={msg.thought_process}
+                                    onChallenge={msg.role === "assistant" ? () => handleChallenge(idx) : undefined}
+                                />
+                            ))}
+                            {isLoading && (
+                                <MessageBubble role="assistant" content="" isLoading={true} />
+                            )}
+                        </div>
+                    )}
+                </div>
+
+                {/* Input Area */}
+                <div className="flex-shrink-0 bg-white pt-10">
+                    <ChatInput
+                        value={query}
+                        onChange={setQuery}
+                        onSubmit={handleSearch}
+                        isLoading={isLoading}
+                        onFileSelect={setAttachedFile}
+                        attachedFile={attachedFile}
+                    />
+                </div>
+            </div>
+        </div>
+    );
+}

frontend/components/Chat/MessageBubble.tsx

@@ -0,0 +1,121 @@
+"use client";
+
+import ReactMarkdown from "react-markdown";
+import { Copy, ThumbsUp, ThumbsDown, User, Bot, Loader2 } from "lucide-react";
+import { SourceList } from "../ui/SourceList";
+import { cleanMessageContent, linkifySources } from "@/lib/utils";
+
+interface MessageBubbleProps {
+    role: "user" | "assistant";
+    content: string;
+    sources?: any[];
+    isLoading?: boolean;
+    intent?: string;
+    thought_process?: string;
+    onChallenge?: () => void;
+}
+
+export default function MessageBubble({ role, content, sources, isLoading, intent, thought_process, onChallenge }: MessageBubbleProps) {
+    const isUser = role === "user";
+
+    // Intent Badge Config
+    const getIntentBadge = (intent?: string) => {
+        if (!intent) return null;
+        if (intent === "SEARCH_REQUIRED") return <span className="text-[10px] uppercase font-bold tracking-wider px-2 py-0.5 rounded-full bg-blue-100 text-blue-700">Web Search</span>;
+        if (intent === "CHAT_ONLY") return <span className="text-[10px] uppercase font-bold tracking-wider px-2 py-0.5 rounded-full bg-emerald-100 text-emerald-700">Direct Chat</span>;
+        if (intent === "CODING_TASK") return <span className="text-[10px] uppercase font-bold tracking-wider px-2 py-0.5 rounded-full bg-purple-100 text-purple-700">Coding</span>;
+        return <span className="text-[10px] uppercase font-bold tracking-wider px-2 py-0.5 rounded-full bg-gray-100 text-gray-700">{intent}</span>;
+    };
+
+    return (
+        <div className={`w-full py-6 ${isUser ? "" : "bg-gray-50 dark:bg-transparent"}`}>
+            <div className="max-w-3xl mx-auto px-4 flex gap-6">
+                {/* Avatar */}
+                <div className="flex-shrink-0 mt-1">
+                    {isUser ? (
+                        <div className="w-8 h-8 rounded-full bg-gray-200 flex items-center justify-center">
+                            <User size={18} className="text-gray-500" />
+                        </div>
+                    ) : (
+                        <div className="w-8 h-8 rounded-full bg-indigo-600 flex items-center justify-center shadow-lg shadow-indigo-500/20">
+                            <Bot size={18} className="text-white" />
+                        </div>
+                    )}
+                </div>
+
+                {/* Content */}
+                <div className="flex-1 overflow-hidden">
+                    <div className="flex items-center gap-3 mb-1">
+                        <span className="font-semibold text-sm text-gray-900">
+                            {isUser ? "You" : "Copilot"}
+                        </span>
+                        {!isUser && !isLoading && getIntentBadge(intent)}
+                    </div>
+
+                    <div className="prose prose-sm max-w-none text-gray-800 leading-relaxed">
+                        {isLoading ? (
+                            <div className="flex items-center gap-2 text-gray-500 animate-pulse">
+                                <Loader2 size={16} className="animate-spin" />
+                                Thinking...
+                            </div>
+                        ) : (
+                            <ReactMarkdown
+                                components={{
+                                    a: ({ node, ...props }) => <a {...props} className="text-blue-600 hover:underline" target="_blank" rel="noopener noreferrer" />
+                                }}
+                            >
+                                {isUser ? content : linkifySources(cleanMessageContent(content), sources)}
+                            </ReactMarkdown>
+                        )}
+                    </div>
+
+                    {/* Sources for Assistant */}
+                    {!isUser && sources && sources.length > 0 && (
+                        <div className="mt-4 pt-4 border-t border-gray-100">
+                            <SourceList sources={sources} />
+                        </div>
+                    )}
+
+                    {/* Reasoning for Assistant (AI-OS Feature) */}
+                    {!isUser && thought_process && (
+                        <div className="mt-4">
+                            <details className="group">
+                                <summary className="flex items-center gap-2 text-xs font-semibold text-gray-400 cursor-pointer hover:text-gray-600 transition-colors select-none">
+                                    <span>🧠 View Thought Process</span>
+                                </summary>
+                                <div className="mt-2 text-xs text-gray-600 font-mono bg-gray-100 p-3 rounded-lg leading-relaxed whitespace-pre-wrap">
+                                    {thought_process}
+                                </div>
+                            </details>
+                        </div>
+                    )}
+
+                    {/* Actions for Assistant */}
+                    {!isUser && !isLoading && (
+                        <div className="flex items-center gap-2 mt-4 text-gray-400">
+                            <button className="p-1 hover:text-gray-600 transition-colors" title="Copy">
+                                <Copy size={14} />
+                            </button>
+                            <button className="p-1 hover:text-gray-600 transition-colors" title="Good response">
+                                <ThumbsUp size={14} />
+                            </button>
+                            <button className="p-1 hover:text-gray-600 transition-colors" title="Bad response">
+                                <ThumbsDown size={14} />
+                            </button>
+                            {/* "Disagree with Me" Button */}
+                            {onChallenge && (
+                                <button
+                                    onClick={onChallenge}
+                                    className="ml-2 flex items-center gap-1 text-xs font-medium text-amber-600 hover:text-amber-700 bg-amber-50 px-2 py-1 rounded border border-amber-200 transition-colors"
+                                    title="Critique this answer"
+                                >
+                                    <span>⚖️ Challenge</span>
+                                </button>
+                            )}
+                        </div>
+                    )}
+                </div>
+            </div>
+        </div>
+    );
+}

frontend/components/Chat/Sidebar.tsx

@@ -0,0 +1,90 @@
+"use client";
+
+import { MessageSquare, Plus, Trash2 } from "lucide-react";
+import { motion } from "framer-motion";
+
+interface SidebarProps {
+    isOpen: boolean;
+    sessions: { id: string; title: string; date: string }[];
+    currentSessionId: string | null;
+    onNewChat: () => void;
+    onSelectSession: (id: string) => void;
+    onDeleteSession: (id: string, e: React.MouseEvent) => void;
+}
+
+export default function Sidebar({
+    isOpen,
+    sessions,
+    currentSessionId,
+    onNewChat,
+    onSelectSession,
+    onDeleteSession,
+}: SidebarProps) {
+    return (
+        <motion.div
+            initial={{ width: 0, opacity: 0 }}
+            animate={{ width: isOpen ? 260 : 0, opacity: isOpen ? 1 : 0 }}
+            className="h-screen bg-gray-50 border-r border-gray-200 flex flex-col flex-shrink-0 overflow-hidden"
+        >
+            {/* New Chat Button */}
+            <div className="p-4">
+                <button
+                    onClick={onNewChat}
+                    className="w-full flex items-center gap-2 px-4 py-3 bg-white hover:bg-gray-100 text-gray-900 rounded-lg transition-colors border border-gray-200 text-sm font-medium shadow-sm"
+                >
+                    <Plus size={16} />
+                    New Chat
+                </button>
+            </div>
+
+            {/* Session List */}
+            <div className="flex-1 overflow-y-auto px-2 pb-4">
+                <div className="text-xs font-semibold text-gray-500 mb-2 px-2 mt-2">
+                    Recent
+                </div>
+                <div className="space-y-1">
+                    {sessions.map((session) => (
+                        <button
+                            key={session.id}
+                            onClick={() => onSelectSession(session.id)}
+                            className={`group w-full flex items-center justify-between gap-2 px-3 py-3 rounded-lg text-sm text-left transition-colors ${currentSessionId === session.id
+                                ? "bg-white text-gray-900 shadow-sm border border-gray-100"
+                                : "text-gray-600 hover:bg-gray-100"
+                                }`}
+                        >
+                            <div className="flex items-center gap-2 overflow-hidden">
+                                <MessageSquare size={14} className="text-gray-400 flex-shrink-0" />
+                                <span className="truncate">{session.title}</span>
+                            </div>
+
+                            <div
+                                onClick={(e) => onDeleteSession(session.id, e)}
+                                className={`text-gray-400 hover:text-red-500 opacity-0 group-hover:opacity-100 transition-opacity p-1`}
+                            >
+                                <Trash2 size={14} />
+                            </div>
+                        </button>
+                    ))}
+                    {sessions.length === 0 && (
+                        <div className="px-4 py-4 text-xs text-gray-400 text-center italic">
+                            No recent chats
+                        </div>
+                    )}
+                </div>
+            </div>
+
+            {/* User Profile / Footer (Optional) */}
+            <div className="p-4 border-t border-gray-200">
+                <div className="flex items-center gap-3 px-2 py-2 text-gray-700 text-sm">
+                    <div className="w-8 h-8 rounded-full bg-indigo-600 flex items-center justify-center text-white font-bold">
+                        U
+                    </div>
+                    <div className="flex-col flex">
+                        <span className="font-medium text-gray-900">User</span>
+                        <span className="text-xs text-gray-500">Pro Plan</span>
+                    </div>
+                </div>
+            </div>
+        </motion.div>
+    );
+}

frontend/components/ui/SearchBar.tsx

@@ -0,0 +1,58 @@
+"use client";
+
+import { Search, ArrowRight } from "lucide-react";
+import { useState } from "react";
+import { cn } from "@/lib/utils";
+
+interface SearchBarProps {
+    onSearch: (query: string) => void;
+    isLoading: boolean;
+    centered?: boolean;
+}
+
+export function SearchBar({ onSearch, isLoading, centered = false }: SearchBarProps) {
+    const [query, setQuery] = useState("");
+
+    const handleSubmit = (e: React.FormEvent) => {
+        e.preventDefault();
+        if (query.trim()) {
+            onSearch(query);
+        }
+    };
+
+    return (
+        <div className={cn("w-full transition-all duration-500", centered ? "max-w-2xl" : "max-w-4xl")}>
+            <form onSubmit={handleSubmit} className="relative group">
+                <div className="absolute inset-y-0 left-4 flex items-center pointer-events-none text-muted-foreground">
+                    <Search className="w-5 h-5" />
+                </div>
+                <input
+                    type="text"
+                    value={query}
+                    onChange={(e) => setQuery(e.target.value)}
+                    placeholder="Ask anything..."
+                    className="w-full py-4 pl-12 pr-12 text-lg bg-secondary/50 border-none rounded-2xl ring-1 ring-black/5 focus:ring-2 focus:ring-primary/20 focus:bg-background transition-all shadow-sm group-hover:shadow-md"
+                    disabled={isLoading}
+                />
+                <button
+                    type="submit"
+                    disabled={!query.trim() || isLoading}
+                    className="absolute inset-y-2 right-2 flex items-center justify-center p-2 rounded-xl bg-primary text-primary-foreground disabled:opacity-50 disabled:cursor-not-allowed hover:opacity-90 transition-opacity"
+                >
+                    {isLoading ? (
+                        <div className="w-5 h-5 border-2 border-current border-t-transparent rounded-full animate-spin" />
+                    ) : (
+                        <ArrowRight className="w-5 h-5" />
+                    )}
+                </button>
+            </form>
+            {centered && (
+                <div className="mt-8 flex gap-2 justify-center text-sm text-muted-foreground">
+                    <span className="px-3 py-1 bg-secondary rounded-full cursor-pointer hover:bg-secondary/80">Market Trends</span>
+                    <span className="px-3 py-1 bg-secondary rounded-full cursor-pointer hover:bg-secondary/80">Code Debugging</span>
+                    <span className="px-3 py-1 bg-secondary rounded-full cursor-pointer hover:bg-secondary/80">Legal Research</span>
+                </div>
+            )}
+        </div>
+    );
+}

frontend/components/ui/SourceList.tsx

@@ -0,0 +1,42 @@
+import { ExternalLink, FileText } from "lucide-react";
+
+interface Source {
+    title: string;
+    url: string;
+    snippet: string;
+}
+
+interface SourceListProps {
+    sources: Source[];
+}
+
+export function SourceList({ sources }: SourceListProps) {
+    if (!sources || sources.length === 0) return null;
+
+    return (
+        <div className="mb-8">
+            <h3 className="text-sm font-semibold text-muted-foreground mb-3 flex items-center gap-2">
+                <FileText className="w-4 h-4" /> Sources
+            </h3>
+            <div className="flex gap-3 overflow-x-auto pb-4 scrollbar-hide">
+                {sources.map((source, idx) => (
+                    <a
+                        key={idx}
+                        href={source.url}
+                        target="_blank"
+                        rel="noopener noreferrer"
+                        className="flex-shrink-0 w-64 p-3 bg-card border rounded-xl hover:bg-secondary/50 transition-colors group cursor-pointer"
+                    >
+                        <div className="text-xs text-muted-foreground mb-1 truncate">{new URL(source.url).hostname}</div>
+                        <div className="font-medium text-sm line-clamp-2 mb-2 group-hover:text-primary transition-colors">
+                            {source.title}
+                        </div>
+                        <div className="text-xs text-muted-foreground line-clamp-2">
+                            {source.snippet}
+                        </div>
+                    </a>
+                ))}
+            </div>
+        </div>
+    );
+}

frontend/env_example.txt

@@ -0,0 +1 @@
+NEXT_PUBLIC_API_BASE=https://YOUR-BACKEND-URL.onrender.com

frontend/eslint.config.mjs

@@ -0,0 +1,18 @@
+import { defineConfig, globalIgnores } from "eslint/config";
+import nextVitals from "eslint-config-next/core-web-vitals";
+import nextTs from "eslint-config-next/typescript";
+
+const eslintConfig = defineConfig([
+  ...nextVitals,
+  ...nextTs,
+  // Override default ignores of eslint-config-next.
+  globalIgnores([
+    // Default ignores of eslint-config-next:
+    ".next/**",
+    "out/**",
+    "build/**",
+    "next-env.d.ts",
+  ]),
+]);
+
+export default eslintConfig;

frontend/lib/utils.ts

@@ -0,0 +1,37 @@
+import { clsx, type ClassValue } from "clsx";
+import { twMerge } from "tailwind-merge";
+
+export function cn(...inputs: ClassValue[]) {
+    return twMerge(clsx(inputs));
+}
+
+export function cleanMessageContent(content: string): string {
+    if (!content) return "";
+
+    // Remove "Assumptions & Risks" section and everything until the next header or end
+    let cleaned = content.replace(/### ⚠️ Assumptions & Risks[\s\S]*?(?=### |$)/g, "");
+
+    // Remove "Unknowns / Limitations" section and everything until the next header or end
+    cleaned = cleaned.replace(/### ❌ Unknowns \/ Limitations[\s\S]*?(?=### |$)/g, "");
+
+    return cleaned.trim();
+}
+
+export function linkifySources(content: string, sources: any[] = []): string {
+    if (!content || !sources.length) return content;
+
+    let linkedContent = content;
+
+    sources.forEach((source, index) => {
+        const sourceNum = index + 1;
+        const url = source.url || "#";
+        // Regex to find "Source X" or "(Source X)" case insensitive
+        // We use a complex regex to avoid double linking if run multiple times or overlapping
+        // Simplified: Look for "Source X" that isn't already inside a markdown link
+        const regex = new RegExp(`(?<!\\[)Source ${sourceNum}(?!\\])`, 'gi');
+
+        linkedContent = linkedContent.replace(regex, `[Source ${sourceNum}](${url})`);
+    });
+
+    return linkedContent;
+}

frontend/next.config.ts

@@ -0,0 +1,7 @@
+import type { NextConfig } from "next";
+
+const nextConfig: NextConfig = {
+  /* config options here */
+};
+
+export default nextConfig;

frontend/package.json

@@ -0,0 +1,32 @@
+{
+  "name": "frontend",
+  "version": "0.1.0",
+  "private": true,
+  "scripts": {
+    "dev": "next dev",
+    "build": "next build",
+    "start": "next start",
+    "lint": "eslint"
+  },
+  "dependencies": {
+    "clsx": "^2.1.1",
+    "framer-motion": "^12.26.2",
+    "lucide-react": "^0.562.0",
+    "next": "16.1.1",
+    "react": "19.2.3",
+    "react-dom": "19.2.3",
+    "react-markdown": "^10.1.0",
+    "tailwind-merge": "^3.4.0",
+    "tailwindcss-animate": "^1.0.7"
+  },
+  "devDependencies": {
+    "@tailwindcss/postcss": "^4",
+    "@types/node": "^20",
+    "@types/react": "^19",
+    "@types/react-dom": "^19",
+    "eslint": "^9",
+    "eslint-config-next": "16.1.1",
+    "tailwindcss": "^4",
+    "typescript": "^5"
+  }
+}

frontend/postcss.config.mjs

@@ -0,0 +1,7 @@
+const config = {
+  plugins: {
+    "@tailwindcss/postcss": {},
+  },
+};
+
+export default config;