Deep Agent – Search + Virtual Files

This repository contains a minimal deep agent implementation based on the "Deep Agents from Scratch" course. The agent can:

1. Search the web (simple Bing wrapper).
2. Create and modify files in an in‑memory virtual file system.
3. Read those virtual files.
4. Dump the virtual file system to the real disk at the end of the run.

Setup

# Create a virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Prerequisite – An Ollama server must be running locally with a model such as llama3.1 available.

Usage

python agent.py

You will be prompted to type questions. The agent will decide which tool to use. When you are finished, type exit.

The virtual files are stored in memory. To persist them, use the DumpVirtualFS tool or call dump_virtual_fs_tool("output") from within the agent.

Files

• agent.py – Main agent implementation.
• virtual_fs.py – Simple in‑memory file system.
• requirements.txt – Python dependencies.
• README.md – This documentation.