Global Hack Week: Agents, Challenge 3

#ai #cli #coding #devchallenge

Day 3: Installing R-CLI (or Backboard Studio) + First Walkthrough

What I Learned

I learned that R-CLI is Backboard’s AI coding agent that runs directly in the terminal.

It doesn’t just autocomplete code — it plans, writes, runs, and even fixes errors recursively by delegating tasks to child agents.

I also discovered that there’s a desktop alternative called Backboard Studio, which provides a VS Code–like interface but uses the same recursive engine.

This challenge taught me that I won’t need to hand-write code in later challenges; instead, I’ll describe what I want and let the agent build it.

What I Did

I set up my environment (on Windows, I installed WSL to get a Linux terminal).
I installed R-CLI using the official install script.
I verified the installation with backboard --version and explored commands with backboard --help.
I logged in using the device code flow, which saved my credentials locally.
I started my first interactive session with backboard.
I tried slash commands like /help, /model, /skills, and /mcp to explore features.
I gave the agent its first job: create a hello.py file that prints a random programming joke, then run it.
I watched R-CLI write the file, run it, and fix errors automatically.
I pushed further by asking it to build a simple index.html countdown website for Global Hack Week, and it delivered.
I noted that I could also use Backboard Studio for the same walkthrough if I preferred a GUI.

Reflection

I realized how powerful recursive agents are compared to simple autocomplete tools.

The ability to write, run, and self-correct code makes R-CLI feel like a true collaborator.

I also appreciated that Studio exists for those who prefer a visual interface, but I enjoyed the raw terminal experience.

This challenge gave me confidence that I can rely on the agent to handle coding tasks in upcoming challenges.