Skip to content

Latest commit

 

History

History
76 lines (55 loc) · 2.72 KB

README.md

File metadata and controls

76 lines (55 loc) · 2.72 KB

Agentory

This repository contains two LLM-powered agent systems for information retrieval and web search. These systems leverage LLMs, semantic search, and web scraping to enhance knowledge discovery and retrieval.

Projects in This Repository

1️⃣ Agentic Retrieval-Augmented Generation (RAG)

A retrieval-augmented generation system that utilizes ChromaDB for efficient document retrieval.

📌 Features:

  • Embeds and indexes documents.
  • Uses ChromaDB for vector-based semantic search.
  • Supports multiple LLM providers: LiteLLM, HfApi, Transformers.
  • Interactive Gradio UI for easy interaction.

📂 Find more details in the Agentic RAG directory.

2️⃣ Multi-Agent Web Search

A multi-agent system that performs web searches and retrieves webpage content for basde on user's query.

📌 Features:

  • Uses DuckDuckGo search API to find relevant web pages.
  • Converts webpage content into Markdown format for readability.
  • A manager agent oversees the web search agent.
  • Supports multiple LLM providers: LiteLLM, HfApi, Transformers.
  • Interactive Gradio UI for seamless user interaction.

📂 Find more details in the Multi-Agent directory.

3️⃣ Browser Automation VLM Agent

A Browser Automation Vision-Language Model (VLM) Agent that automates web browsing tasks, such as navigating web pages, searching for content, handling popups, and extracting information.

📌 Features:

  • Web Automation: Navigate to web pages, click on elements, and search within pages.
  • Popup Handling: Close modals and popups with ease.
  • Information Extraction: Extract and process webpage content.
  • Screenshot Capture: Automatically captures screenshots during interactions.
  • LLM Support: Works with models from LiteLLM, HfApi, and Transformers.
  • Gradio UI: Interactive web interface for user interaction.

📂 Find more details in the Browser Automation directory.

Quick Start

  • Clone the repo.
  • uv is recommended for easy install and setup for all agents in this repo. You only need run each script in their directory (uv will create the venv and install all the dependencies for you).

Agentic RAG

cd agentic_rag
uv run agentic_rag_chromadb.py

Multi-Agent Web Search

cd multi_agent
uv run multi_agent_web_search.py

Browser Automation VLM Agent

cd browser_auto_agent
uv run browser_auto_vlm_agent.py

License

See LICENSE for more details.

Contributing

Contributions are welcome! Feel free to open issues and submit pull requests. 🚀