camel-ai/owl ? reverse-engineered prompt

I want to build a system in Python that uses a team of AI agents to automate complex tasks. The main idea is to create an "optimized workforce" of agents. There should be a primary manager agent that can take a high-level goal, break it down into smaller, actionable steps, and then delegate those steps to a team of specialized worker agents.

These worker agents need to be able to interact with the real world. They should have tools to browse the web, perform searches using different engines, read and write files locally, and run terminal commands. The goal is to create a robust multi-agent framework that can solve real-world problems.

Please also create a simple web interface where I can input a task and watch the agents work, seeing their plan and progress. Make it flexible so I can plug in different LLMs, like models from OpenAI, Gemini, or open-source ones via platforms like OpenRouter.