tools

Why we included this project

Agents that need to actually touch the world usually end up calling the same handful of actions: reading and writing files, running shell commands, storing memory, calling APIs, and doing math. Strands Agents Tools bundles these as ready-made capabilities you can attach to an agent at initialization, then either let the model pick the right one or call them directly as methods. Custom tools are just a decorator away, and anything you drop into a ./tools directory gets loaded automatically and reloaded when you edit it, which saves a lot of fuss while iterating. If you are already on the Strands Python or TypeScript SDK, this is the quickest way to give an agent real filesystem access instead of leaving it stuck in text. It is a library rather than an app, so you will want it alongside an agent framework or the MCP server.