Multi-Agent Research Planner

A minimal multi-agent research workflow using OpenAI and Brave Search.

This project demonstrates a simple agent orchestration pattern in Python:

• ResearchPlannerAgent gathers a research plan from the user and saves it via database tools.
• WebSearchAgent derives search terms from the plan and queries the Brave Search API.
• SummaryReportAgent summarizes the collected search results into Markdown.

Files

• main.py — application entrypoint and agent definitions.
• database.py — database initialization and persistence helpers.
• README.md — this documentation.

Requirements

• Python 3.11+
• openai Python package
• python-dotenv
• pydantic
• requests

Setup

1. Create a virtual environment and activate it:

python -m venv .venv
.\.venv\Scripts\Activate.ps1

2. Install dependencies:

pip install openai python-dotenv pydantic requests

3. Create a .env file in the project root with your API keys:

OPENAI_API_KEY=your_openai_api_key
BRAVE_API_KEY=your_brave_search_api_key

Usage

Run the application from the repository root:

python main.py

The program will prompt for a research task. You can type input to build the plan interactively.

Special commands during the prompt loop:

• exit — quit the program
• accept — finalize the research plan and continue to web search and summarization

After completion, a summary_report.md file is written with the final summary.

Notes

• WebSearchAgent uses the Brave Search API and expects BRAVE_API_KEY to be set.
• The current flow is simple and can be extended with additional tools, more agents, or improved prompt handling.