strands-agents/tools ? reverse-engineered prompt

Build me a Python package for agent tools that gives an AI assistant a bunch of practical abilities in one place.

I want it to support things like reading and writing files, making HTTP requests, running shell commands, doing math and Python code execution, and a few richer agent workflows like web search, browser actions, Slack access, memory storage, task scheduling, and connecting to external MCP tools. It should feel easy to plug into an agent, with a clean way to enable only the tools I want and examples that show how to use them.

Please include solid docs, simple tests, and a good default structure so it’s easy to extend with new tools later. A few of the more powerful tools should be clearly marked as experimental or unsafe, with obvious warnings and safety checks where needed. If you need current library details, look up the latest docs online before you implement anything.