Skip to content

USER_GUIDE.md

Latest commit

 

History

History
121 lines (90 loc) · 3.61 KB

USER_GUIDE.md

File metadata and controls

121 lines (90 loc) · 3.61 KB
Raw

📘 Sentinel User Guide

Where can Sentinel be used?

Sentinel is designed to be a versatile component in your AI stack. You can use it in three main ways:

1. As a Python Library (pip install sentinel-core)

Integrate Sentinel directly into your Python applications (FastAPI, Django, Flask, or scripts).

  • Use case: Building a custom RAG pipeline, a research assistant, or a data monitoring bot.
  • How: Import Sentinel, GraphManager, and GraphExtractor to build custom workflows.

2. As a CLI Tool

Run Sentinel from your terminal to manage knowledge graphs without writing code.

  • Use case: Quick scraping, monitoring a specific URL, or checking system status.
  • How: Run sentinel init, sentinel watch <url>, sentinel status.

3. As a RAG Knowledge Source

Use the Neo4j graph created by Sentinel as a rich context source for your LLMs.

  • Use case: Chatbots that need historical context or structured data (not just vector similarity).
  • How: Connect your RAG chain (LangChain, LlamaIndex) to the Neo4j database populated by Sentinel.

🚀 How to Use Sentinel

Prerequisite: Infrastructure

Sentinel requires:

  1. Neo4j Database: To store the knowledge graph.
  2. LLM: Ollama (local) or OpenAI/Anthropic (cloud) to extract knowledge.

Method 1: The "No-Code" CLI Approach

Best for getting started quickly.

  1. Install:

    pip install sentinel-core

  2. Initialize:

    # Sets up your .env file and checks connections
python -m sentinel_core.cli init

  3. Watch a URL:

    # Scrapes, extracts, and saves to Neo4j
python -m sentinel_core.cli watch https://example.com

  4. Auto-Heal:

    # Checks for stale data and updates it
python -m sentinel_core.cli heal

Method 2: The Python API Approach

Best for building custom applications.

import asyncio
from sentinel_core import Sentinel, GraphManager, GraphExtractor
from sentinel_core.scraper import get_scraper

async def main():
    # 1. Setup components
    graph = GraphManager()  # Connects to Neo4j
    scraper = get_scraper() # Uses Firecrawl or Local
    extractor = GraphExtractor(model_name="ollama/phi3") # Uses your LLM
    
    sentinel = Sentinel(graph, scraper, extractor)
    
    # 2. Process a website
    print("Processing...")
    result = await sentinel.process_url("https://docs.python.org/3/")
    
    print(f"Done! Extracted {result['extracted_nodes']} nodes.")
    
    # 3. Query the graph (Time Travel)
    # Get the state of the graph as it was yesterday
    # snapshot = graph.get_graph_snapshot(timestamp=yesterday)

if __name__ == "__main__":
    asyncio.run(main())

Method 3: Integrating with LangChain (RAG)

Once Sentinel has populated your Neo4j database, you can query it using LangChain.

from langchain_community.graphs import Neo4jGraph
from langchain.chains import GraphCypherQAChain
from langchain_openai import ChatOpenAI

# 1. Connect to the same Neo4j instance Sentinel is using
graph = Neo4jGraph(
    url="bolt://localhost:7687", 
    username="neo4j", 
    password="password"
)

# 2. Create a QA chain
chain = GraphCypherQAChain.from_llm(
    ChatOpenAI(temperature=0), 
    graph=graph, 
    verbose=True
)

# 3. Ask questions based on Sentinel's data
response = chain.run("What entities are related to 'Python'?")
print(response)

📚 Further Reading