Every developer knows the feeling: you have a great idea for a command-line tool or a simple script to automate a tedious task, but the thought of setting up the boilerplate, parsing arguments, and creating the file structure is the definition of dev toil.
But what if you could just describe the tool you want, hand an AI the documentation, and have it build the tool for you?
In the short video below, I demonstrate a new open-source project that does exactly that. I'm using Amp, the AI-native terminal, to create a "tool-generator" for its own Toolbox feature.
Watch as I give Amp a single prompt:
I provide the URL to the Amp Toolbox documentation.
I ask it to set up a repository in such a way that any user can describe a new tool, and Amp will automatically generate it based on the docs.
Amp autonomously reads the documentation, understands the requirements, scaffolds an entire directory structure with templates and setup scripts, and hands me back a ready-to-use "Toolbox Playground." To prove it works, I then ask the new system to "Run Python tests in a project directory," and it instantly creates a brand new, functional tool for me.
This is the type of work to give to an AI Coding Agent: implementation details for things you’ve already thought through. You do the thinking, let the AI do the “typing”.
Check out the GitHub repo here: https://github.com/DannyMac180/amp-toolbox-playground
Learn more about the Amp Toolbox: https://ampcode.com/news/toolboxes
