feat: add Gemini support, fix Neo4j Aura connection, and add setup guide

Author: aditya452007

SETUP_GUIDE.md

@@ -0,0 +1,182 @@
+# 🚀 Secrin Setup Guide
+
+Welcome to Secrin! This guide covers two ways to run the project:
+
+1. **☁️ Hybrid Mode (Recommended):** Cloud DB + Gemini API. Fast, lightweight, runs on any laptop.  
+2. **🔒 Local Mode:** Local Docker DB + Ollama. Privacy-focused, requires good hardware (16GB+ RAM).
+
+---
+
+## 🛠️ 1. Prerequisites (Check these first!)
+Open your terminal and check these versions. If missing, install them.
+
+- **Git:** `git --version`  
+- **Python (3.10+):** `python --version`  
+- **Node.js (v20+):** `node -v`  
+- **Poetry:** `pip install poetry`  
+- **(Optional) Docker:** Required only if you choose "Local Mode".  
+- **(Optional) Ollama:** Required only if you choose "Local Mode".
+
+---
+
+## 🔑 2. Credentials & Services
+Choose **Option A** (Easier) or **Option B** (Private).
+
+### Option A — Hybrid (Cloud DB + Gemini)
+1. **Database:** Go to https://console.neo4j.io  
+    - Create a **Free Instance**.  
+    - ⚠️ **Copy the password immediately.** You won't see it again.  
+    - Copy the URI (example: `neo4j+s://abc12345.databases.neo4j.io`).  
+2. **AI Model:** Get a key from Google AI Studio: https://aistudio.google.com/app/apikey
+
+### Option B — Fully Local (Docker + Ollama)
+1. **Database:** Run this Docker command to start Neo4j locally:
+```bash
+docker run --restart always --publish=7474:7474 --publish=7687:7687 --env NEO4J_AUTH=neo4j/password neo4j:latest
+```
+2. **AI Model:** Install Ollama and pull the model:
+```bash
+ollama pull llama3.2
+ollama serve
+```
+
+---
+
+## 🖥️ 3. Backend Setup (Terminal 1)
+
+### Step 3.1 — Install dependencies
+```bash
+git clone https://github.com/SecrinLabs/secrin.git
+cd secrin
+poetry install
+```
+
+### Step 3.2 — Configure environment (.env)
+Create the .env file from the example:
+
+```bash
+cp .env.example .env
+```
+
+Open `.env` and paste the configuration matching your choice from Step 2.
+
+Option A (Hybrid / Gemini) — example `.env`:
+
+```Bash
+# --- Database (Neo4j Aura) ---
+NEO4J_URI=neo4j+s://<YOUR_INSTANCE_ID>.databases.neo4j.io
+NEO4J_USER=neo4j
+NEO4J_PASS=<YOUR_SAVED_PASSWORD>
+NEO4J_DB=neo4j
+
+# --- Embedding (Local CPU/GPU) ---
+EMBEDDING_PROVIDER=sentence_transformer
+SENTENCE_TRANSFORMER_MODEL=sentence-transformers/all-MiniLM-L6-v2
+EMBEDDING_DIMENSION=384
+
+# --- LLM (Gemini) ---
+LLM_PROVIDER=gemini
+GOOGLE_API_KEY=<YOUR_GEMINI_KEY>
+```
+
+Option B (Local / Ollama) — example `.env`:
+
+```Bash
+# --- Database (Local Docker) ---
+NEO4J_URI=bolt://localhost:7687
+NEO4J_USER=neo4j
+NEO4J_PASS=password
+NEO4J_DB=neo4j
+
+# --- Embedding (Ollama) ---
+EMBEDDING_PROVIDER=ollama
+OLLAMA_BASE_URL=http://localhost:11434
+OLLAMA_EMBEDDING_MODEL=mxbai-embed-large
+EMBEDDING_DIMENSION=1024
+
+# --- LLM (Ollama) ---
+LLM_PROVIDER=ollama
+LLM_MODEL_OLLAMA=llama3.2
+```
+
+### Step 3.3 — Launch backend
+```bash
+# 1. Verify connection
+poetry run setup
+
+# 2. Start server
+poetry run server
+```
+Success: you should see Uvicorn running on http://0.0.0.0:8000.
+
+---
+
+## 🎨 4. Frontend Setup (Terminal 2)
+
+Open a new terminal window:
+```bash
+cd secrin/apps/web
+```
+
+### Step 4.1 — Install dependencies
+The project prefers pnpm. Try these methods in order.
+
+Method 1 — Standard:
+```bash
+corepack enable
+pnpm install
+```
+
+Method 2 — If you get "Permission Denied" (Windows) (PowerShell):
+
+```powershell
+iwr https://get.pnpm.io/install.ps1 -useb | iex
+```
+Then retry Method 1.
+
+Method 3 — Fallback to npm:
+```bash
+npm install
+```
+
+### Step 4.2 — Start UI
+```bash
+cp .env.example .env
+pnpm dev   # OR 'npm run dev' if you used Method 3
+```
+Success: open http://localhost:3000
+
+---
+
+## 🚀 5. First Run & Ingestion
+- The graph will be blank on first login. This is normal.  
+- Login using your Neo4j credentials.  
+- Hybrid users: if `neo4j+s://` fails to connect, try `bolt+s://`.
+
+Ingest data:
+1. Go to "Integrations" in the sidebar.  
+2. Paste a GitHub URL (example: `https://github.com/SecrinLabs/secrin`).  
+3. Click "Ingest".  
+4. Watch the backend terminal — it will clone the repo and embed it.  
+5. When the backend terminal says "Finished", go to the Graph tab to view your data.
+
+---
+
+## ❓ Troubleshooting — Common Errors
+
+- "No routing servers available"  
+  1. Check if Neo4j Aura is "Paused" in the console. Resume it.  
+  2. Try using `bolt+s://` instead of `neo4j+s://`.  
+  3. Try a mobile hotspot (some Wi‑Fi blocks port 7687).
+
+- "Database does not exist"  
+  - Neo4j Aura Free tier only allows the database name `neo4j`. Check your `.env`.
+
+- "Command pnpm not found"  
+  - Restart your terminal after installing it. If it still fails, in PowerShell run:
+```powershell
+$env:PATH = "$env:LOCALAPPDATA\pnpm;$env:PATH"
+```
+
+- "Unsupported LLM provider: gemini"  
+  - You are running an old version. `git pull` the latest changes or check `llm_factory.py`.

apps/web/components/Neo4jGraph.tsx

@@ -92,11 +92,17 @@ export default function Neo4jGraph({
       // Dynamic import of neo4j-driver for client-side only
       const neo4j = (await import("neo4j-driver")).default;
 
+      // FIX: Check if the URL implies a secure connection (Cloud/Aura)
+      const isSecure = neo4jUrl.includes('+s');
+
       const driver = neo4j.driver(
         neo4jUrl,
         neo4j.auth.basic(username, password),
-        { encrypted: false }
+        // If it's a secure URL, don't pass any config (let the URL handle it).
+        // If it's local, explicitly turn off encryption.
+        isSecure ? {} : { encrypted: "ENCRYPTION_OFF" }
       );
+      
       const session = driver.session({ database });
 
       const start = Date.now();

packages/config/settings.py

@@ -100,6 +100,9 @@ class Settings(BaseSettings):
         description="Embedding cache TTL in seconds (if caching enabled)"
     )
 
+    # Google Api Configuration
+    GOOGLE_API_KEY: Optional[str] = None
+
     # OpenAI Configuration
     OPENAI_API_KEY: str = Field(
         default="",

packages/memory/factories/llm_factory.py

@@ -6,6 +6,8 @@
 from typing import Optional
 from packages.memory.llm.base import BaseLLMProvider
 from packages.memory.llm.providers.ollama import OllamaProvider
+# --- ADDED: Import Gemini ---
+from packages.memory.llm.providers.gemini import GeminiProvider
 from packages.config import Settings
 
 
@@ -47,10 +49,16 @@ def create_provider(
                 temperature=temperature,
                 max_tokens=max_tokens
             )
+        elif provider.lower() == "gemini":
+            return GeminiProvider(
+                model=model,
+                temperature=temperature,
+                max_tokens=max_tokens
+            )
         else:
             raise ValueError(
                 f"Unsupported LLM provider: {provider}. "
-                f"Currently supported: ['ollama']"
+                f"Currently supported: ['ollama', 'gemini']"
             )
 
     @classmethod

packages/memory/llm/providers/gemini.py

@@ -0,0 +1,88 @@
+"""
+Gemini LLM provider implementation.
+"""
+
+import requests
+import json
+import os
+from typing import List, Any, Optional, Iterator
+from packages.memory.llm.base import BaseLLMProvider
+from packages.config.settings import Settings
+
+class GeminiProvider(BaseLLMProvider):
+    """Gemini provider using Google's Generative Language API."""
+
+    def __init__(
+        self,
+        model: Optional[str] = "gemini-2.5-flash",
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        api_key: Optional[str] = None
+    ):
+        settings = Settings()
+        
+        # 1. Configuration
+        model = model or "gemini-2.5-flash"
+        temperature = temperature if temperature is not None else settings.LLM_TEMPERATURE
+        max_tokens = max_tokens or settings.LLM_MAX_TOKENS
+        
+        super().__init__(model=model, temperature=temperature, max_tokens=max_tokens)
+
+        # 2. Auth: Try settings first, then fallback to OS environment
+        self.api_key = api_key or getattr(settings, "GOOGLE_API_KEY", None) or os.getenv("GOOGLE_API_KEY")
+        
+        if not self.api_key:
+            raise ValueError("GOOGLE_API_KEY is missing. Please add it to your .env file.")
+
+    def get_provider_name(self) -> str:
+        return "gemini"
+
+    def is_available(self) -> bool:
+        return bool(self.api_key)
+
+    def generate_answer(self, question: str, context_items: List[Any], search_type: str) -> str:
+        """Generate answer using Gemini."""
+        prompt = self._build_prompt(question, context_items, search_type)
+        return self.generate_text(prompt)
+
+    def generate_text(self, prompt: str, system_prompt: Optional[str] = None) -> str:
+        """
+        Generate text using Gemini REST API.
+        """
+        url = f"https://generativelanguage.googleapis.com/v1beta/models/{self.model}:generateContent?key={self.api_key}"
+        
+        # Combine system prompt if exists (Gemini supports system_instruction but simple concatenation is robust)
+        full_prompt = prompt
+        if system_prompt:
+            full_prompt = f"System: {system_prompt}\n\nUser: {prompt}"
+
+        payload = {
+            "contents": [{
+                "parts": [{"text": full_prompt}]
+            }],
+            "generationConfig": {
+                "temperature": self.temperature,
+                "maxOutputTokens": self.max_tokens
+            }
+        }
+
+        try:
+            response = requests.post(url, json=payload, headers={"Content-Type": "application/json"}, timeout=30)
+            response.raise_for_status()
+            
+            data = response.json()
+            # Extract text from complex Gemini JSON response
+            try:
+                return data["candidates"][0]["content"]["parts"][0]["text"]
+            except (KeyError, IndexError):
+                return ""
+                
+        except Exception as e:
+            raise Exception(f"Gemini API Error: {str(e)}")
+
+    def stream_text(self, prompt: str, system_prompt: Optional[str] = None) -> Iterator[str]:
+        """
+        Stream text using Gemini (Simplified to non-streaming for stability first).
+        """
+        # For the initial setup, we simply yield the full result to avoid async complexities
+        yield self.generate_text(prompt, system_prompt)