v1

.gitignore

@@ -1,3 +1,18 @@
 .env
-venv/*
-data/*
+.env.*
+!.env.example
+
+venv/
+.venv/
+env/
+
+__pycache__/
+*.py[cod]
+
+# Local backups / temp
+legal_agentic_system.git.backup/
+
+# Optional large local data (kept out of Git by default)
+legal_agentic_system/data/
+legal_agentic_system/faiss_index/
+legal_agentic_system/policy_faiss_index/

README.md

@@ -0,0 +1,149 @@
+# Citizen Navigator AI Portal
+
+A dual-dashboard Flask + LangGraph application for Indian citizen assistance:
+- **⚖️ Legal Resolution Dashboard**: upload a notice/document or describe a case and get official-step guidance.
+- **🎓 Opportunity Finder Dashboard**: submit a detailed profile and get personalized schemes, jobs, innovation grants, missed opportunities, and a practical roadmap.
+
+The system uses **hybrid retrieval** for profile intelligence:
+- **Live web search** (official domains via Tavily)
+- **Offline policy RAG** (FAISS over local policy PDFs)
+
+---
+
+## Features
+
+- Two independent pipelines (no cross-contamination of legal/profile workflows)
+- Stage-aware legal analysis (`legal_category`, `case_status`)
+- Deep profile filtering (`gender`, `income`, `status`, `skills`, `goal`)
+- Structured profile outputs:
+  - `opportunities_json`
+  - `missed_opportunities_json`
+  - `guiding_roadmap`
+- Dark-mode, dual-tab frontend with async API calls
+- OCR fallback for scanned PDF/image documents
+
+---
+
+## Project Structure
+
+```text
+legal/
+├── README.md
+├── .gitignore
+├── legal_agentic_system/
+│   ├── app.py
+│   ├── main_graph.py
+│   ├── state.py
+│   ├── requirements.txt
+│   ├── templates/
+│   │   └── index.html
+│   ├── agents/
+│   │   ├── data_agent.py
+│   │   ├── legal_pipeline.py
+│   │   ├── profile_pipeline.py
+│   │   ├── policy_rag.py
+│   │   └── search_tool.py
+│   └── ...
+└── venv/ (local)
+```
+
+---
+
+## Architecture Overview
+
+### 1) Legal Graph
+`data_extraction -> legal_analysis -> END`
+
+- `data_extraction`: reads raw text / text file / PDF / image OCR
+- `legal_analysis`: summarizes issue, performs official procedure search, returns legal action plan
+
+### 2) Profile Graph
+`profile_analysis -> END`
+
+- `profile_analysis`: combines policy RAG + live search, ranks opportunities, flags missed windows, builds roadmap
+
+---
+
+## Requirements
+
+- Python 3.10+
+- Poppler installed (required by `pdf2image`)
+- API keys for Groq and Tavily
+
+Install dependencies:
+
+```bash
+cd legal_agentic_system
+python -m venv .venv
+source .venv/bin/activate
+pip install -r requirements.txt
+```
+
+---
+
+## Environment Variables
+
+Create `legal_agentic_system/.env`:
+
+```env
+GROQ_API_KEY=your_groq_key
+TAVILY_API_KEY=your_tavily_key
+HF_TOKEN=your_huggingface_token
+```
+
+Optional policy-RAG corpus:
+- Put policy PDFs in: `legal_agentic_system/policy_docs/`
+- FAISS index is auto-created in: `legal_agentic_system/policy_faiss_index/`
+
+---
+
+## Run Locally
+
+```bash
+cd legal_agentic_system
+python app.py
+```
+
+Open: `http://localhost:5000`
+
+---
+
+## API Endpoints
+
+- `POST /api/analyze_legal`
+  - Inputs: `raw_text` **or** `file`, plus optional `legal_category`, `case_status`
+  - Output: `solution`
+
+- `POST /api/analyze_profile`
+  - Inputs: profile fields (`user_age`, `user_state`, `user_education`, `user_specialization`, `user_gender`, `annual_income`, `current_status`, `extracurriculars`, `user_goal`)
+  - Output: `opportunities`, `missed_opportunities`, `roadmap`, `report`
+
+---
+
+## Why Files May Not Have Appeared on GitHub
+
+This repository previously had `legal_agentic_system` tracked as a **nested git link** instead of normal files. That setup was converted so the folder contents are now tracked directly by this repo.
+
+After pulling these changes, run:
+
+```bash
+git add -A
+git commit -m "Fix tracking + add README"
+git push
+```
+
+---
+
+## Notes
+
+- Large/generated directories are intentionally gitignored by default:
+  - `legal_agentic_system/data/`
+  - `legal_agentic_system/faiss_index/`
+  - `legal_agentic_system/policy_faiss_index/`
+- If you want these committed, remove those lines from `.gitignore`.
+
+---
+
+## License
+
+Use according to your organization/course requirements.

legal_agentic_system

@@ -1 +0,0 @@
-Subproject commit cde3d5423892cec9fcf998c4def1e9a9698b7070

legal_agentic_system/.gitignore

@@ -0,0 +1,51 @@
+# Environment / secrets
+.env
+.env.*
+!.env.example
+
+# Python bytecode and caches
+__pycache__/
+*.py[cod]
+*$py.class
+
+# Virtual environments
+.venv/
+venv/
+env/
+ENV/
+
+# Build / packaging artifacts
+build/
+dist/
+*.egg-info/
+.eggs/
+pip-wheel-metadata/
+
+# Test / type-check / tooling caches
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+.coverage
+.coverage.*
+htmlcov/
+
+# IDE / OS files
+.vscode/
+.idea/
+.DS_Store
+Thumbs.db
+
+# Logs
+*.log
+
+# Runtime / temp files
+tmp/
+temp/
+temp_page_*.jpg
+
+# Local vector DB / retrieval artifacts
+faiss_index/
+policy_faiss_index/
+
+# Local app data (optional large/generated assets)
+data/

legal_agentic_system/agents/data_agent.py

@@ -0,0 +1,103 @@
+import os
+import base64
+from dotenv import load_dotenv
+from huggingface_hub import InferenceClient
+from pdf2image import convert_from_path
+from pypdf import PdfReader
+from state import LegalGraphState
+
+# Load environment variables
+load_dotenv()
+
+def encode_image_to_base64(image_path):
+    with open(image_path, "rb") as image_file:
+        return base64.b64encode(image_file.read()).decode('utf-8')
+
+def process_document(state: LegalGraphState):
+    # 1. Check if raw text was passed directly
+    if state.get("raw_text") and not state.get("file_path"):
+        print("---RECEIVED DIRECT TEXT INPUT---")
+        return {"raw_text": state["raw_text"], "current_step": "summarization"}
+        
+    file_path = state.get("file_path", "")
+    if not file_path or not os.path.exists(file_path):
+        raise ValueError("No valid file path or raw text provided.")
+
+    # 2. Handle Plain Text Files
+    if file_path.lower().endswith('.txt'):
+        print("---READING PLAIN TEXT FILE---")
+        with open(file_path, 'r', encoding='utf-8') as f:
+            extracted_text = f.read().strip()
+        return {"raw_text": extracted_text, "current_step": "summarization"}
+
+    # 3. Handle PDFs (Try PyPDF first, fallback to OCR if empty/scanned)
+    if file_path.lower().endswith('.pdf'):
+        print("---ATTEMPTING FAST TEXT EXTRACTION VIA PYPDF---")
+        try:
+            reader = PdfReader(file_path)
+            extracted_text = ""
+            for page in reader.pages:
+                text = page.extract_text()
+                if text:
+                    extracted_text += text + "\n"
+            
+            # If PyPDF finds a meaningful amount of text, return it immediately
+            if len(extracted_text.strip()) > 50:
+                print("---SUCCESSFULLY EXTRACTED TEXT VIA PYPDF---")
+                return {"raw_text": extracted_text.strip(), "current_step": "summarization"}
+            else:
+                print("---PYPDF FOUND NO TEXT (LIKELY SCANNED). FALLING BACK TO OCR---")
+        except Exception as e:
+            print(f"---PYPDF ERROR: {e}. FALLING BACK TO OCR---")
+
+    # 4. Fallback: Hugging Face Vision OCR (For Images and Scanned PDFs)
+    print("---EXTRACTING TEXT VIA HUGGING FACE VISION API---")
+    hf_token = os.getenv("HF_TOKEN")
+    if not hf_token:
+        raise ValueError("HF_TOKEN is missing. Please check your .env file.")
+        
+    client = InferenceClient(token=hf_token)
+    extracted_text = ""
+    image_paths = []
+
+    if file_path.lower().endswith(('.png', '.jpg', '.jpeg')):
+        image_paths.append(file_path)
+    elif file_path.lower().endswith('.pdf'):
+        # Convert the scanned PDF to images
+        pages = convert_from_path(file_path, dpi=150)
+        for i, page in enumerate(pages):
+            temp_path = f"temp_page_{i}.jpg"
+            page.save(temp_path, 'JPEG')
+            image_paths.append(temp_path)
+    else:
+        raise ValueError(f"Unsupported file type: {file_path}")
+
+    # Process images through the Vision Model
+    for img_path in image_paths:
+        img_b64 = encode_image_to_base64(img_path)
+        
+        messages = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": "You are a legal OCR system. Extract all readable text from this document image exactly as written. Do not add any commentary."},
+                    {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{img_b64}"}}
+                ]
+            }
+        ]
+        
+        try:
+            response = client.chat_completion(
+                model="meta-llama/Llama-3.2-11B-Vision-Instruct",
+                messages=messages,
+                max_tokens=1500
+            )
+            extracted_text += response.choices[0].message.content + "\n\n"
+        except Exception as e:
+            extracted_text += f"[Error processing image: {e}]\n"
+        finally:
+            # Clean up generated image pages from PDFs
+            if file_path.lower().endswith('.pdf') and os.path.exists(img_path):
+                os.remove(img_path)
+
+    return {"raw_text": extracted_text.strip(), "current_step": "summarization"}

legal_agentic_system/agents/legal_agent.py

@@ -0,0 +1,39 @@
+import os
+
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_groq import ChatGroq
+from state import LegalGraphState
+from rag_setup import constitution_retriever
+from dotenv import load_dotenv
+load_dotenv()
+
+groq_api_key = os.getenv("GROQ_API_KEY")
+
+llm = ChatGroq(temperature=0.2, model_name="llama-3.1-8b-instant")
+
+def suggest_legal_solution(state: LegalGraphState):
+    summary = state["summary"]
+    
+    retrieved_docs = constitution_retriever.invoke(summary)
+    rag_context = "\n".join([doc.page_content for doc in retrieved_docs])
+    
+    prompt = ChatPromptTemplate.from_messages([
+    ("system", """You are an elite Legal Strategist and Expert Counsel in Indian Law. 
+Your objective is to analyze the provided case summary alongside the retrieved legal context to formulate a highly accurate, practical, and objective legal solution.
+
+The case may involve constitutional rights, civil disputes, criminal offenses, corporate law, or administrative issues. Adapt your analysis to fit the specific facts and context provided.
+
+You MUST structure your response strictly using the following headings:
+1. 📌 Key Legal Issues: (Identify the core disputes, violations, or questions of law)
+2. ⚖️ Applicable Law & Interpretation: (Explain exactly how the provided legal context applies to the case)
+3. 🛠️ Suggested Recourse & Strategy: (Provide actionable, step-by-step legal remedies or defense strategies)
+4. ⚠️ Risk Assessment & Caveats: (Briefly highlight any legal risks, burdens of proof, or exceptions)
+
+Maintain a professional, authoritative, and objective legal tone. Do not invent laws; rely strictly on the provided context and established Indian legal principles."""),
+    ("human", "Case Summary:\n{summary}\n\nRetrieved Legal Context:\n{context}\n\nGenerate the structured legal analysis:")
+])
+    
+    chain = prompt | llm
+    response = chain.invoke({"summary": summary, "context": rag_context})
+    
+    return {"rag_context": rag_context, "legal_solution": response.content, "current_step": "opportunities"}

legal_agentic_system/agents/legal_pipeline.py

@@ -0,0 +1,60 @@
+import os
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_groq import ChatGroq
+from state import CitizenState
+from agents.search_tool import gov_search_tool
+
+llm = ChatGroq(
+    temperature=0.1,
+    model_name="llama-3.1-8b-instant",
+    api_key=os.getenv("GROQ_API_KEY"),
+)
+
+
+def analyze_legal_case(state: CitizenState):
+    category = state.get("legal_category", "General")
+    status = state.get("case_status", "Unknown Stage")
+    raw_text = state.get("raw_text", "")
+
+    summary_prompt = ChatPromptTemplate.from_messages([
+        ("system", "Summarize the core legal issue or dispute from the text in 3 concise bullet points."),
+        ("human", "{text}"),
+    ])
+    summary = (summary_prompt | llm).invoke({"text": (raw_text or "")[:10000]}).content
+
+    search_query = (
+        f"Government of India official procedure for {category} dispute "
+        f"at stage: {status}. Case details: {summary}"
+    )
+    live_data = gov_search_tool.invoke({"query": search_query})
+
+    solution_prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """You are a Citizen Legal Advisor. Based on the user's issue and the live government data, provide a structured legal action plan.
+        Do not hallucinate laws. Only use the provided live data. Include website links if present in the data.
+        Format cleanly with markdown headings:
+        - ⚖️ Case Assessment
+        - 🛠️ Recommended Official Steps
+        - 🏛️ Official Portals / Resources""",
+        ),
+        (
+            "human",
+            "Category: {category}\nStage: {status}\nUser Issue: {summary}\n\nLive Gov Data: {live_data}",
+        ),
+    ])
+
+    solution = (solution_prompt | llm).invoke(
+        {
+            "category": category,
+            "status": status,
+            "summary": summary,
+            "live_data": str(live_data),
+        }
+    ).content
+
+    return {
+        "legal_summary": summary,
+        "legal_solution": solution,
+        "gov_resources": str(live_data),
+    }

legal_agentic_system/agents/policy_rag.py

@@ -0,0 +1,39 @@
+import os
+from langchain_community.document_loaders import DirectoryLoader, PyPDFLoader
+from langchain_text_splitters import RecursiveCharacterTextSplitter
+from langchain_huggingface import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+
+
+def get_policy_retriever(db_path: str = "policy_faiss_index"):
+    embeddings = HuggingFaceEmbeddings(
+        model_name="sentence-transformers/all-MiniLM-L6-v2",
+        model_kwargs={"device": "cpu"},
+    )
+
+    if os.path.exists(db_path):
+        return FAISS.load_local(
+            db_path,
+            embeddings,
+            allow_dangerous_deserialization=True,
+        ).as_retriever(search_kwargs={"k": 4})
+
+    docs_dir = "policy_docs"
+    if not os.path.exists(docs_dir):
+        return None
+
+    loader = DirectoryLoader(docs_dir, glob="**/*.pdf", loader_cls=PyPDFLoader)
+    docs = loader.load()
+    if not docs:
+        return None
+
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=300)
+    splits = text_splitter.split_documents(docs)
+
+    vectorstore = FAISS.from_documents(splits, embeddings)
+    vectorstore.save_local(db_path)
+
+    return vectorstore.as_retriever(search_kwargs={"k": 4})
+
+
+policy_retriever = get_policy_retriever()

legal_agentic_system/agents/profile_pipeline.py

@@ -0,0 +1,169 @@
+import os
+import json
+import re
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_groq import ChatGroq
+from state import CitizenState
+from agents.search_tool import gov_search_tool
+from agents.policy_rag import policy_retriever
+
+llm = ChatGroq(
+    temperature=0.1,
+    model_name="llama-3.3-70b-versatile",
+    api_key=os.getenv("GROQ_API_KEY"),
+)
+
+
+def _extract_json_block(text: str):
+    try:
+        return json.loads(text)
+    except Exception:
+        pass
+
+    match = re.search(r"\[.*\]", text, re.DOTALL)
+    if match:
+        try:
+            return json.loads(match.group(0))
+        except Exception:
+            return []
+
+    return []
+
+
+def analyze_profile_opportunities(state: CitizenState):
+    age = str(state.get("user_age") or "").strip()
+    gender = str(state.get("user_gender") or "").strip()
+    income = str(state.get("annual_income") or "").strip()
+    user_status = str(state.get("current_status") or "").strip()
+    edu = str(state.get("user_education") or "").strip()
+    spec = str(state.get("user_specialization") or "").strip()
+    loc = str(state.get("user_state") or "").strip()
+    extra = str(state.get("extracurriculars") or "").strip()
+    goal = str(state.get("user_goal") or "Find best suited opportunities").strip()
+
+    profile_context = (
+        f"Age: {age}, Gender: {gender}, Income: {income}, Status: {user_status}, "
+        f"Education: {edu}, Specialization: {spec}, Location: {loc}, Experience: {extra}, Goal: {goal}"
+    )
+
+    rag_query = (
+        f"Eligibility and rules for {user_status or 'student'} in {loc or 'India'} "
+        f"regarding {goal}. Consider gender {gender} and income {income}."
+    )
+    rag_context = ""
+    if policy_retriever is not None:
+        rag_docs = policy_retriever.invoke(rag_query)
+        rag_context = "\n".join([doc.page_content for doc in rag_docs])
+
+    active_query = (
+        f"Government of India active schemes, grants, and jobs for {gender} {user_status}, "
+        f"{edu} in {spec}, income {income}, located in {loc}"
+    )
+    active_data = gov_search_tool.invoke({"query": active_query})
+
+    innovation_query = (
+        f"Government of India startup grants, innovation funds, or fellowships "
+        f"for students with experience in: {extra}"
+    )
+    innovation_data = gov_search_tool.invoke({"query": innovation_query})
+
+    missed_query = (
+        f"Recently closed government deadlines, expired schemes, or age limits passed "
+        f"for age {age} {spec} graduates in India"
+    )
+    missed_data = gov_search_tool.invoke({"query": missed_query})
+
+    opportunities_prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """You are an elite AI Citizen Advisor.
+Return ONLY valid JSON array with exactly 3 objects. Do not include markdown or prose.
+Each object must contain:
+- name
+- category
+- why_this_suits_you
+- eligibility_fit
+- average_time_to_redeem
+- accessibility_score
+- action_steps (array of short strings)
+- official_link
+
+Rules:
+1) Use only provided RAG context and live government data.
+2) Filter out schemes/jobs the user is unlikely to qualify for (age, gender, income, status).
+3) Keep output strict JSON.""",
+        ),
+        (
+            "human",
+            "Profile: {profile}\n\nPolicy RAG Context:\n{rag_context}\n\nActive Data: {active}\n\nInnovation Data: {innovation}",
+        ),
+    ])
+
+    opportunities_raw = (opportunities_prompt | llm).invoke(
+        {
+            "profile": profile_context,
+            "rag_context": rag_context,
+            "active": str(active_data),
+            "innovation": str(innovation_data),
+        }
+    ).content
+
+    opportunities_json = _extract_json_block(opportunities_raw)
+
+    missed_prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """Return ONLY valid JSON array of up to 3 missed opportunities.
+Each object must include:
+- name
+- reason_missed
+- next_cycle_hint
+- expected_next_window
+- official_link""",
+        ),
+        (
+            "human",
+            "Profile: {profile}\n\nMissed Data: {missed}",
+        ),
+    ])
+
+    missed_raw = (missed_prompt | llm).invoke(
+        {
+            "profile": profile_context,
+            "missed": str(missed_data),
+        }
+    ).content
+    missed_json = _extract_json_block(missed_raw)
+
+    roadmap_prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """Create a concise, actionable roadmap in markdown.
+Sections:
+### 🎯 Core Eligibility Schemes
+### 🚀 Career & Innovation Grants
+### ⚠️ Missed Opportunities
+Use empathetic plain language and include practical next actions.""",
+        ),
+        (
+            "human",
+            "Profile: {profile}\n\nSelected Opportunities: {opps}\n\nMissed: {missed}",
+        ),
+    ])
+
+    roadmap = (roadmap_prompt | llm).invoke(
+        {
+            "profile": profile_context,
+            "opps": json.dumps(opportunities_json, ensure_ascii=False),
+            "missed": json.dumps(missed_json, ensure_ascii=False),
+        }
+    ).content
+
+    return {
+        "active_opportunities": str(active_data),
+        "missed_opportunities": str(missed_data),
+        "opportunities_json": opportunities_json,
+        "missed_opportunities_json": missed_json,
+        "guiding_roadmap": roadmap,
+        "insight_report": roadmap,
+    }

legal_agentic_system/agents/recommendation_agent.py

@@ -0,0 +1,66 @@
+import os
+from dotenv import load_dotenv
+from langchain_community.tools.tavily_search import TavilySearchResults
+from langchain_groq import ChatGroq
+from langchain_core.prompts import ChatPromptTemplate
+from state import LegalGraphState
+
+load_dotenv()
+groq_api_key = os.getenv("GROQ_API_KEY")
+tavily_api_key = os.getenv("TAVILY_API_KEY")
+
+llm = ChatGroq(temperature=0.3, model_name="llama-3.1-8b-instant", api_key=groq_api_key)
+search_tool = TavilySearchResults(max_results=4, tavily_api_key=tavily_api_key)
+
+def suggest_opportunities(state: LegalGraphState):
+    age = str(state.get("user_age") or "Unknown").strip()
+    edu = str(state.get("user_education") or "Unknown").strip()
+    spec = str(state.get("user_specialization") or "Unknown").strip()
+    loc = str(state.get("user_state") or "India").strip()
+
+    if edu == "Unknown" and spec == "Unknown":
+        return {
+            "recommended_policies": "",
+            "recommended_exams_jobs": "",
+            "current_step": "report_generation",
+        }
+
+    policy_query = f"Government of India schemes, scholarships, or policies for {edu} students in {spec} from {loc}"
+    policy_results = search_tool.invoke({"query": policy_query})
+
+    career_query = f"Upcoming government exams and tech jobs in India for {spec} graduates age {age}"
+    career_results = search_tool.invoke({"query": career_query})
+
+    prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """You are a helpful Career and Policy Advisor for Indian citizens.
+        Based on the user's profile and the live search results, format two clean, distinct sections:
+        1. **Beneficial Government Policies & Schemes** (Include actionable steps and website links if available).
+        2. **Relevant Government Exams & Job Opportunities** (Tailored to their education and field).
+
+        Keep the tone encouraging, accessible, and completely free of AI or technical jargon.""",
+        ),
+        (
+            "human",
+            "User Profile: Age {age}, Education: {edu} in {spec}, Location: {loc}\n\nPolicy Data: {policy_data}\n\nCareer Data: {career_data}",
+        ),
+    ])
+
+    chain = prompt | llm
+    response = chain.invoke(
+        {
+            "age": age,
+            "edu": edu,
+            "spec": spec,
+            "loc": loc,
+            "policy_data": str(policy_results),
+            "career_data": str(career_results),
+        }
+    )
+
+    return {
+        "recommended_policies": response.content,
+        "recommended_exams_jobs": response.content,
+        "current_step": "report_generation",
+    }

legal_agentic_system/agents/report_agent.py

@@ -0,0 +1,48 @@
+import os
+from dotenv import load_dotenv
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_groq import ChatGroq
+from state import LegalGraphState
+
+load_dotenv()
+groq_api_key = os.getenv("GROQ_API_KEY")
+llm = ChatGroq(temperature=0.1, model_name="llama-3.1-8b-instant", api_key=groq_api_key)
+
+def compile_final_report(state: LegalGraphState):
+    summary = state.get("summary", "")
+    solution = state.get("legal_solution", "")
+    policies = state.get("recommended_policies", "")
+    exams_jobs = state.get("recommended_exams_jobs", "")
+
+    prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            """You are an expert public servant and citizen advocate. Your job is to compile the provided legal analysis and policy recommendations into one comprehensive, easy-to-read Final Report.
+
+        Rules:
+        1. STRIP ALL JARGON: Do not use terms like "LLM", "RAG", "Agent", "Context", or "Prompt".
+        2. Use clear headings, bullet points, and bold text for readability.
+        3. Structure the report as follows:
+           - 📝 Case Overview (Simplify the legal summary)
+           - ⚖️ Actionable Legal Guidance (Simplify the legal solution)
+           - 🎓 Citizen Opportunities (Display the policies, exams, and jobs)
+           - 🌐 Next Steps & Support Links
+        4. Speak directly to the user in a helpful, empathetic, and professional tone.""",
+        ),
+        (
+            "human",
+            "Legal Summary: {summary}\n\nLegal Solution: {solution}\n\nPolicies: {policies}\n\nExams and Jobs: {exams_jobs}",
+        ),
+    ])
+
+    chain = prompt | llm
+    response = chain.invoke(
+        {
+            "summary": summary,
+            "solution": solution,
+            "policies": policies,
+            "exams_jobs": exams_jobs,
+        }
+    )
+
+    return {"final_report": response.content, "current_step": "end"}

legal_agentic_system/agents/resource_agent.py

@@ -0,0 +1,29 @@
+import os
+
+from langchain_community.tools.tavily_search import TavilySearchResults
+from langchain_groq import ChatGroq
+from langchain_core.prompts import ChatPromptTemplate
+from state import LegalGraphState
+from dotenv import load_dotenv
+load_dotenv()
+
+groq_api_key = os.getenv("GROQ_API_KEY")
+
+llm = ChatGroq(temperature=0, model_name="llama-3.3-70b-versatile")
+search_tool = TavilySearchResults(max_results=3)
+
+def find_gov_resources(state: LegalGraphState):
+    solution = state["legal_solution"]
+    
+    search_query = f"official government of India portal or scheme for: {solution[:100]}"
+    search_results = search_tool.invoke({"query": search_query})
+    
+    prompt = ChatPromptTemplate.from_messages([
+        ("system", "You are an administrative assistant. Format the following web search results into a clean list of actionable government steps, resources, and official website links. Ensure links are included if available."),
+        ("human", "Search Results: {results}")
+    ])
+    
+    chain = prompt | llm
+    response = chain.invoke({"results": str(search_results)})
+    
+    return {"government_resources": [response.content], "current_step": "end"}

legal_agentic_system/agents/search_tool.py

@@ -0,0 +1,13 @@
+import os
+from dotenv import load_dotenv
+from langchain_community.tools.tavily_search import TavilySearchResults
+
+load_dotenv()
+tavily_api_key = os.getenv("TAVILY_API_KEY")
+
+gov_search_tool = TavilySearchResults(
+    max_results=4,
+    search_depth="advanced",
+    include_domains=["gov.in", "nic.in", "india.gov.in"],
+    tavily_api_key=tavily_api_key,
+)

legal_agentic_system/agents/summary_agent.py

@@ -0,0 +1,25 @@
+import os
+
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_groq import ChatGroq
+from state import LegalGraphState
+from dotenv import load_dotenv
+load_dotenv()
+
+groq_api_key = os.getenv("GROQ_API_KEY")
+
+llm = ChatGroq(temperature=0, model_name="llama-3.3-70b-versatile")
+
+def summarize_document(state: LegalGraphState):
+    raw_text = state["raw_text"]
+    
+    prompt = ChatPromptTemplate.from_messages([
+        ("system", "You are an expert legal clerk. Extract the core legal dispute, parties involved, and key facts from the following text. Keep it objective and concise."),
+        ("human", "Legal Document: \n\n{text}")
+    ])
+    
+    chain = prompt | llm
+    # Truncate text slightly to ensure it fits in the context window
+    response = chain.invoke({"text": raw_text[:20000]}) 
+    
+    return {"summary": response.content, "current_step": "legal_analysis"}

legal_agentic_system/app.py

@@ -0,0 +1,88 @@
+import os
+import tempfile
+from flask import Flask, request, render_template, jsonify
+from werkzeug.utils import secure_filename
+from dotenv import load_dotenv
+from main_graph import legal_graph, profile_graph
+
+load_dotenv()
+
+app = Flask(__name__)
+app.secret_key = os.urandom(24)
+
+ALLOWED_EXTENSIONS = {'pdf', 'png', 'jpg', 'jpeg', 'txt'}
+
+def allowed_file(filename):
+    return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS
+
+@app.route('/')
+def index():
+    return render_template('index.html')
+
+
+@app.route('/api/analyze_legal', methods=['POST'])
+def analyze_legal():
+    raw_text = request.form.get('raw_text', '').strip()
+    inputs = {}
+    temp_path = None
+
+    inputs["legal_category"] = request.form.get('legal_category', 'General')
+    inputs["case_status"] = request.form.get('case_status', 'Need Information')
+
+    if raw_text:
+        inputs["raw_text"] = raw_text
+        inputs["file_path"] = None
+    elif 'file' in request.files and request.files['file'].filename != '':
+        file = request.files['file']
+        if not allowed_file(file.filename):
+            return jsonify({"success": False, "error": "Invalid file type. Please upload a PDF, TXT, JPG, or PNG."}), 400
+
+        filename = secure_filename(file.filename)
+        temp_path = os.path.join(tempfile.gettempdir(), filename)
+        file.save(temp_path)
+        inputs["file_path"] = temp_path
+        inputs["raw_text"] = None
+    else:
+        return jsonify({"success": False, "error": "No file or text provided."}), 400
+
+    try:
+        final_state = legal_graph.invoke(inputs)
+        return jsonify({
+            "success": True,
+            "solution": final_state.get("legal_solution", "Analysis failed."),
+        })
+    except Exception as e:
+        return jsonify({"success": False, "error": str(e)}), 500
+    finally:
+        if temp_path and os.path.exists(temp_path):
+            os.remove(temp_path)
+
+
+@app.route('/api/analyze_profile', methods=['POST'])
+def analyze_profile():
+    inputs = {
+        "user_age": request.form.get('user_age', ''),
+        "user_state": request.form.get('user_state', ''),
+        "user_education": request.form.get('user_education', ''),
+        "user_specialization": request.form.get('user_specialization', ''),
+        "user_gender": request.form.get('user_gender', ''),
+        "annual_income": request.form.get('annual_income', ''),
+        "current_status": request.form.get('current_status', ''),
+        "extracurriculars": request.form.get('extracurriculars', ''),
+        "user_goal": request.form.get('user_goal', ''),
+    }
+
+    try:
+        final_state = profile_graph.invoke(inputs)
+        return jsonify({
+            "success": True,
+            "opportunities": final_state.get("opportunities_json", []),
+            "missed_opportunities": final_state.get("missed_opportunities_json", []),
+            "roadmap": final_state.get("guiding_roadmap", ""),
+            "report": final_state.get("insight_report", "Report generation failed."),
+        })
+    except Exception as e:
+        return jsonify({"success": False, "error": str(e)}), 500
+
+if __name__ == '__main__':
+    app.run(debug=True, host='0.0.0.0', port=5000)

legal_agentic_system/main_graph.py

@@ -0,0 +1,31 @@
+from langgraph.graph import StateGraph, END
+from state import CitizenState
+from agents.data_agent import process_document
+from agents.legal_pipeline import analyze_legal_case
+from agents.profile_pipeline import analyze_profile_opportunities
+
+# --- GRAPH 1: Legal Dashboard ---
+def build_legal_graph():
+    workflow = StateGraph(CitizenState)
+    workflow.add_node("data_extraction", process_document)
+    workflow.add_node("legal_analysis", analyze_legal_case)
+
+    workflow.set_entry_point("data_extraction")
+    workflow.add_edge("data_extraction", "legal_analysis")
+    workflow.add_edge("legal_analysis", END)
+
+    return workflow.compile()
+
+
+# --- GRAPH 2: Profile Dashboard ---
+def build_profile_graph():
+    workflow = StateGraph(CitizenState)
+    workflow.add_node("profile_analysis", analyze_profile_opportunities)
+
+    workflow.set_entry_point("profile_analysis")
+    workflow.add_edge("profile_analysis", END)
+
+    return workflow.compile()
+
+legal_graph = build_legal_graph()
+profile_graph = build_profile_graph()

legal_agentic_system/rag_setup.py

@@ -0,0 +1,31 @@
+import os
+from langchain_community.document_loaders import PyPDFLoader
+from langchain_text_splitters import RecursiveCharacterTextSplitter
+from langchain_huggingface import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+
+def initialize_vector_store(pdf_path="data/indian_constitution.pdf", db_path="faiss_index"):
+    embeddings = HuggingFaceEmbeddings(
+        model_name="sentence-transformers/all-MiniLM-L6-v2",
+        model_kwargs={'device': 'cpu'}
+    )
+    
+    if os.path.exists(db_path):
+        return FAISS.load_local(db_path, embeddings, allow_dangerous_deserialization=True)
+    
+    if not os.path.exists(pdf_path):
+        raise FileNotFoundError(f"Please place the constitution PDF at {pdf_path}")
+        
+    loader = PyPDFLoader(pdf_path)
+    docs = loader.load()
+    
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+    splits = text_splitter.split_documents(docs)
+    
+    vectorstore = FAISS.from_documents(splits, embeddings)
+    vectorstore.save_local(db_path)
+    return vectorstore
+
+# Export a retriever instance
+vectorstore = initialize_vector_store()
+constitution_retriever = vectorstore.as_retriever(search_kwargs={"k": 3})

legal_agentic_system/requirements.txt

@@ -0,0 +1,16 @@
+Flask
+Werkzeug
+huggingface_hub
+pdf2image
+Pillow
+langchain
+langchain-groq
+langchain-community
+langchain-huggingface
+langchain-text-splitters
+langgraph
+tavily-python
+faiss-cpu
+python-dotenv
+pypdf
+sentence-transformer

legal_agentic_system/state.py

@@ -0,0 +1,37 @@
+from typing import TypedDict, Optional
+
+
+class CitizenState(TypedDict):
+    # --- Legal Dashboard Inputs ---
+    file_path: Optional[str]
+    raw_text: Optional[str]
+    legal_category: Optional[str]
+    case_status: Optional[str]
+
+    # --- Profile Dashboard Inputs ---
+    user_age: Optional[str]
+    user_state: Optional[str]
+    user_education: Optional[str]
+    user_specialization: Optional[str]
+    user_gender: Optional[str]
+    annual_income: Optional[str]
+    current_status: Optional[str]
+    extracurriculars: Optional[str]
+    user_goal: Optional[str]
+
+    # --- Legal Outputs ---
+    legal_summary: str
+    legal_solution: str
+    gov_resources: str
+
+    # --- Profile Outputs ---
+    active_opportunities: str
+    missed_opportunities: str
+    opportunities_json: list[dict]
+    missed_opportunities_json: list[dict]
+    guiding_roadmap: str
+    insight_report: str
+
+
+# Backward compatibility for existing modules still importing LegalGraphState
+LegalGraphState = CitizenState

legal_agentic_system/templates/index.html

@@ -0,0 +1,227 @@
+<!DOCTYPE html>
+<html lang="en" class="light">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Citizen Navigator | Dual Dashboard</title>
+    <script src="https://cdn.tailwindcss.com"></script>
+    <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css">
+    <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script>
+    <script>
+        tailwind.config = { darkMode: 'class' }
+    </script>
+    <style>
+        .markdown-body h3 { font-size: 1.25rem; font-weight: 700; margin-top: 1.5em; margin-bottom: 0.75em; border-bottom: 1px solid #e2e8f0; padding-bottom: 0.3em; }
+        .markdown-body p { margin-bottom: 1em; line-height: 1.6; }
+        .markdown-body ul { margin-left: 1.5em; margin-bottom: 1em; list-style-type: disc; }
+        .markdown-body a { color: #2563eb; text-decoration: underline; font-weight: 500; }
+        .dark .markdown-body h3 { border-color: #334155; color: white; }
+        .dark .markdown-body p, .dark .markdown-body li { color: #cbd5e1; }
+        .dark .markdown-body a { color: #60a5fa; }
+    </style>
+</head>
+<body class="bg-slate-50 dark:bg-slate-900 min-h-screen text-slate-800 dark:text-slate-200 transition-colors">
+
+    <nav class="bg-white dark:bg-slate-800 border-b border-slate-200 dark:border-slate-700 px-6 py-4 sticky top-0 z-10">
+        <div class="max-w-7xl mx-auto flex flex-col sm:flex-row items-center justify-between gap-4">
+            <div class="flex items-center gap-3">
+                <div class="bg-blue-600 text-white p-2 rounded-lg"><i class="fa-solid fa-building-columns"></i></div>
+                <h1 class="text-xl font-bold">Citizen Navigator</h1>
+            </div>
+
+            <div class="flex bg-slate-100 dark:bg-slate-700 p-1 rounded-lg">
+                <button onclick="switchTab('legal')" id="btn-legal" class="px-4 py-2 text-sm font-medium rounded-md bg-white dark:bg-slate-800 shadow-sm transition-all">⚖️ Legal Resolution</button>
+                <button onclick="switchTab('profile')" id="btn-profile" class="px-4 py-2 text-sm font-medium rounded-md text-slate-500 dark:text-slate-400 hover:text-slate-700 transition-all">🎓 Opportunity Finder</button>
+            </div>
+
+            <button onclick="document.documentElement.classList.toggle('dark')" class="p-2 rounded-full hover:bg-slate-100 dark:hover:bg-slate-700">
+                <i class="fa-solid fa-circle-half-stroke"></i>
+            </button>
+        </div>
+    </nav>
+
+    <main class="max-w-5xl mx-auto px-6 py-8">
+
+        <div id="view-legal" class="block space-y-8">
+            <div class="bg-white dark:bg-slate-800 rounded-2xl p-6 shadow-sm border border-slate-200 dark:border-slate-700">
+                <h2 class="text-lg font-bold mb-4">Analyze Legal Issue</h2>
+                <form id="legal-form" class="space-y-4">
+                    <div class="grid grid-cols-2 gap-4 mb-4">
+                        <div>
+                            <label class="block text-sm font-medium mb-1">Domain</label>
+                            <select name="legal_category" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                                <option value="Cyber Crime">Cyber Crime</option>
+                                <option value="Tax & Finance">Tax & Finance</option>
+                                <option value="Civil Dispute">Civil Dispute</option>
+                                <option value="Corporate/Startup">Corporate/Startup</option>
+                            </select>
+                        </div>
+                        <div>
+                            <label class="block text-sm font-medium mb-1">Current Stage</label>
+                            <select name="case_status" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                                <option value="Need Information">Just need information</option>
+                                <option value="Received Notice">Received a Government/Legal Notice</option>
+                                <option value="Active Case">Active Court Case</option>
+                            </select>
+                        </div>
+                    </div>
+                    <div>
+                        <label class="block text-sm font-medium mb-1">Upload Notice/Document</label>
+                        <input type="file" name="file" accept=".pdf,.png,.jpg,.txt" class="w-full text-sm border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                    </div>
+                    <div>
+                        <label class="block text-sm font-medium mb-1">Or Describe Situation</label>
+                        <textarea name="raw_text" rows="3" class="w-full border p-3 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600" placeholder="E.g., I received a tax notice..."></textarea>
+                    </div>
+                    <button type="submit" id="btn-submit-legal" class="w-full bg-blue-600 text-white py-3 rounded-xl font-medium hover:bg-blue-700">Find Legal Recourse</button>
+                </form>
+            </div>
+
+            <div id="legal-loading" class="hidden text-center py-10 animate-pulse">Querying official government databases...</div>
+            <div id="legal-results" class="hidden bg-white dark:bg-slate-800 rounded-2xl p-8 shadow-sm border border-slate-200 dark:border-slate-700">
+                <div id="legal-output" class="markdown-body"></div>
+            </div>
+        </div>
+
+        <div id="view-profile" class="hidden space-y-8">
+            <div class="bg-white dark:bg-slate-800 rounded-2xl p-6 shadow-sm border border-slate-200 dark:border-slate-700">
+                <h2 class="text-lg font-bold mb-4">Discover Schemes & Missed Opportunities</h2>
+                <form id="profile-form" class="space-y-4">
+                    <div class="grid grid-cols-2 gap-4">
+                        <div>
+                            <label class="block text-sm font-medium mb-1">Age</label>
+                            <input type="number" name="user_age" placeholder="20" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                        </div>
+                        <div>
+                            <label class="block text-sm font-medium mb-1">State</label>
+                            <input type="text" name="user_state" placeholder="e.g., Uttar Pradesh" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                        </div>
+                    </div>
+                    <div>
+                        <label class="block text-sm font-medium mb-1">Education Level</label>
+                        <input type="text" name="user_education" placeholder="e.g., 3rd Year B.Tech" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                    </div>
+                    <div>
+                        <label class="block text-sm font-medium mb-1">Specialization</label>
+                        <input type="text" name="user_specialization" placeholder="e.g., CS (AIML & IoT)" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                    </div>
+                    <div class="grid grid-cols-2 gap-4 mt-4">
+                        <div>
+                            <label class="block text-sm font-medium mb-1">Gender</label>
+                            <select name="user_gender" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                                <option value="Male">Male</option>
+                                <option value="Female">Female</option>
+                                <option value="Other">Other</option>
+                            </select>
+                        </div>
+                        <div>
+                            <label class="block text-sm font-medium mb-1">Annual Family Income</label>
+                            <select name="annual_income" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                                <option value="Below 2.5 Lakhs">Below 2.5 Lakhs</option>
+                                <option value="2.5 to 8 Lakhs">2.5 to 8 Lakhs</option>
+                                <option value="Above 8 Lakhs">Above 8 Lakhs</option>
+                            </select>
+                        </div>
+                    </div>
+                    <div class="mt-4">
+                        <label class="block text-sm font-medium mb-1">Current Status</label>
+                        <select name="current_status" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                            <option value="Student">Student</option>
+                            <option value="Unemployed">Unemployed</option>
+                            <option value="Working Professional">Working Professional</option>
+                            <option value="Entrepreneur">Entrepreneur</option>
+                        </select>
+                    </div>
+                    <div class="mt-4">
+                        <label class="block text-sm font-medium mb-1">Primary Goal (Optional)</label>
+                        <input type="text" name="user_goal" placeholder="e.g., Start an AI company" class="w-full border p-2 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600">
+                    </div>
+                    <div class="mt-4">
+                        <label class="block text-sm font-medium mb-1">Projects, Leadership, or Skills (Optional)</label>
+                        <textarea name="extracurriculars" rows="2" class="w-full border p-3 rounded-lg bg-slate-50 dark:bg-slate-700 dark:border-slate-600" placeholder="E.g., Built machine learning apps, led a university tech club..."></textarea>
+                    </div>
+                    <button type="submit" id="btn-submit-profile" class="w-full bg-teal-600 text-white py-3 rounded-xl font-medium hover:bg-teal-700">Generate Insight Report</button>
+                </form>
+            </div>
+
+            <div id="profile-loading" class="hidden text-center py-10 animate-pulse">Analyzing active and expired schemes...</div>
+            <div id="profile-results" class="hidden bg-white dark:bg-slate-800 rounded-2xl p-8 shadow-sm border border-slate-200 dark:border-slate-700">
+                <div id="profile-opportunities" class="space-y-3 mb-6"></div>
+                <div id="profile-missed" class="space-y-3 mb-6"></div>
+                <div id="profile-output" class="markdown-body"></div>
+            </div>
+        </div>
+
+    </main>
+
+    <script>
+        function switchTab(tab) {
+            const isLegal = tab === 'legal';
+
+            document.getElementById('view-legal').classList.toggle('hidden', !isLegal);
+            document.getElementById('view-legal').classList.toggle('block', isLegal);
+
+            document.getElementById('view-profile').classList.toggle('hidden', isLegal);
+            document.getElementById('view-profile').classList.toggle('block', !isLegal);
+
+            const btnClassActive = "px-4 py-2 text-sm font-medium rounded-md bg-white dark:bg-slate-800 shadow-sm transition-all text-slate-800 dark:text-white";
+            const btnClassInactive = "px-4 py-2 text-sm font-medium rounded-md text-slate-500 dark:text-slate-400 hover:text-slate-700 transition-all bg-transparent shadow-none";
+
+            document.getElementById('btn-legal').className = isLegal ? btnClassActive : btnClassInactive;
+            document.getElementById('btn-profile').className = !isLegal ? btnClassActive : btnClassInactive;
+        }
+
+        async function handleForm(e, endpoint, btnId, loadingId, resultsId, outputId) {
+            e.preventDefault();
+            const btn = document.getElementById(btnId);
+            const loading = document.getElementById(loadingId);
+            const results = document.getElementById(resultsId);
+
+            btn.disabled = true;
+            btn.innerText = 'Processing...';
+            results.classList.add('hidden');
+            loading.classList.remove('hidden');
+
+            try {
+                const response = await fetch(endpoint, { method: 'POST', body: new FormData(e.target) });
+                const data = await response.json();
+
+                if (data.success) {
+                    if (endpoint === '/api/analyze_profile') {
+                        const oppsBox = document.getElementById('profile-opportunities');
+                        const missedBox = document.getElementById('profile-missed');
+                        const opps = Array.isArray(data.opportunities) ? data.opportunities : [];
+                        const missed = Array.isArray(data.missed_opportunities) ? data.missed_opportunities : [];
+
+                        oppsBox.innerHTML = opps.length
+                            ? '<h3 class="text-lg font-semibold mb-2">🎯 Recommended Opportunities</h3>' + opps.map(item => `\n                                <div class="border border-slate-200 dark:border-slate-700 rounded-lg p-4">\n                                    <div class="font-semibold">${item.name || 'Opportunity'}</div>\n                                    <div class="text-sm text-slate-500 dark:text-slate-400">${item.category || ''}</div>\n                                    <div class="text-sm mt-2">${item.why_this_suits_you || ''}</div>\n                                    <div class="text-xs mt-2 text-slate-500 dark:text-slate-400">Eligibility: ${item.eligibility_fit || 'N/A'} | Time: ${item.average_time_to_redeem || 'N/A'} | Accessibility: ${item.accessibility_score || 'N/A'}</div>\n                                </div>`).join('')
+                            : '';
+
+                        missedBox.innerHTML = missed.length
+                            ? '<h3 class="text-lg font-semibold mb-2">⚠️ Missed Opportunities</h3>' + missed.map(item => `\n                                <div class="border border-amber-200 dark:border-amber-700 rounded-lg p-4">\n                                    <div class="font-semibold">${item.name || 'Opportunity'}</div>\n                                    <div class="text-sm mt-1">${item.reason_missed || ''}</div>\n                                    <div class="text-sm mt-1 text-slate-500 dark:text-slate-400">Next: ${item.next_cycle_hint || ''} (${item.expected_next_window || 'TBD'})</div>\n                                </div>`).join('')
+                            : '';
+
+                        const roadmapText = data.roadmap || data.report || '';
+                        document.getElementById(outputId).innerHTML = marked.parse(roadmapText);
+                    } else {
+                        const outputText = data.solution || data.report;
+                        document.getElementById(outputId).innerHTML = marked.parse(outputText);
+                    }
+                    results.classList.remove('hidden');
+                } else {
+                    alert("Error: " + data.error);
+                }
+            } catch (err) {
+                alert("Server connection failed.");
+            } finally {
+                btn.disabled = false;
+                btn.innerText = btnId.includes('legal') ? 'Find Legal Recourse' : 'Generate Insight Report';
+                loading.classList.add('hidden');
+            }
+        }
+
+        document.getElementById('legal-form').addEventListener('submit', (e) => handleForm(e, '/api/analyze_legal', 'btn-submit-legal', 'legal-loading', 'legal-results', 'legal-output'));
+        document.getElementById('profile-form').addEventListener('submit', (e) => handleForm(e, '/api/analyze_profile', 'btn-submit-profile', 'profile-loading', 'profile-results', 'profile-output'));
+    </script>
+</body>
+</html>