Meet E.V.A My Own Ai CLI Assistant - Fun AI project to learn from

“ Hello Chris. What are we doing today?.”

---

Features

• Web Scraper: Ingests clean, readable content from any URL.
• Local Knowledge Base: Embeds documents using LangChain + OllamaEmbeddings + ChromaDB.
• Smart Query Engine: RAG & agent-based querying with tool invocation.
• Integrated Tools:
  • nmap scanner
  • ping connectivity check
  • find file locator
  • semantic knowledge search
• Memory-Enhanced Agent: React-based LangChain agent with conversational memory.
• Interactive Mode: Stay in the terminal and interact in real-time.

---

Stack

• Python 3.10+
• LangChain
• Ollama (local LLM runner)
• [llama3.1:8b] (current running this model, great for technical assistance)
• Chroma (vector DB)
• BeautifulSoup (scraping)
• .env for config
• Modular tool handler using LangChain’s Tool wrapper

1. Install Dependencies

pip install -r requirements.txt

Ensure ollama is installed and running locally:

ollama run llama3

2. Create a .env file from the example.env

name= (this is where the LLM will refrence you)

Use Cases

Take advanatge of the RAG by injesting documention via URL

python main.py --ingest "https://example.com"

Ask it a direct question 

python main.py --ask "example question"

Interactve mode

Interact with the LLM

python main.py

Then type away

> scan my network
> ping google.com
> find /etc/passwd

How It Works

• RAG Pipeline for standard Q&A

• ReAct Agent invokes tools when needed

• Tool Usage triggered via natural language

• Memory System stores context (last 5 messages)

• Streaming Handler gives live feedback from LLM

• Embeddings generated via OllamaEmbeddings("llama3.1:8b")

Roadmap

[] PDF / Markdown / TXT ingestion

[] GUI wrapper (Gradio / TUI) for desktop companion option

[] Voice command support

[] Docker packaging

Developer Notes

Personalize your assistant:

Swap llama3.1:8b with any local Ollama-compatible LLM

Drop in new Tool functions via LangChain's API