I’m really embracing agentic programming. The coding models have gotten good enough and I’ve obtained enough experience to know what sort of tools to give my agent access to so it can check its work when it stumbles. Bonus: I’m using PeonPing/peon-ping to help me maintain multiple agents while making programming even more like an RTS. So fun!
Without further adieu here’s my list
Treemand: builder for CLI commands
Okay I cannot state how surprised I was the evening I thought up this tool. I thought “surly someone has built this before” but evidently not. Treemand is a CLI tool that lets you explore and build commands from other CLI tools. I and many other programmers usually explore CLI tools by running --help, iteratively discovering the right sub command we care about and ensuring we’re using it right. Treemand affords a TUI to do this exploration in a more visual manner, even allowing you to build up a command that can either be copied or executed.
You can also simply get a print out of the CLI as modeled into as a graph.
Anthem World
I realized I’m not very familiar with most other nation’s National Anthems. When were they written and by who? What are the original lyrics and what do they mean? When did the nation adopt the anthem? What does it sound like?
Inspired by the recent 2026 winter olympics, I am in process of putting together a website that lets you explore a map of the world, clicking on countries to view information on their national anthem. Users can also view a table of this information to easily discover which country has the youngest national anthem or to listen to anthems in regions.
For the laughs I have a “hot or not” style game that users can play and cast votes to contribute to a leaderboard.
Assgen
Giggle. Assgen is a CLI client/server that (1) models the typical 3D video game work that happens as a CLI graph and (2) for each video game development tasks (such as create concept art based on a prompt, or write up a scenario given facts of a world, or create a sound effect) to utilize an open source model (found on hugging face) to complete the task.
At this point there are machine learning models for anything and everything. I figure might as well making utilizing them easier.
The CLI implements a client/server because I want to run the models on my beefy desktop computer while using my sleek laptop.
Users can configure what open source model they wish to use for what video game development task by editing a default configuration. Obviously the user can check their configuration to ensure a user can quickly detect if they’ve put in an invalid model.
Conclusion
I work on these projects and some others during hobbyist time. As a software engineer it has been a blast putting together these projects. I have so many ideas and I had to triage more aggressively where I put my energy in the pre-agentic world. Now I can explore.
What a time to be a programmer.