This is a submission for the GitHub Copilot CLI Challenge
What I Built
Way Finder is an anti‑streak, anti‑gacha habit‑tracking prototype that grew out of losing our cat Stela and realizing how much of my time was going into chasing virtual rewards instead of real life.
I want an app that helps you see your life clearly, track real project progress, build good habits, instantly recall your wins, and still get gentle reminders to slow down and enjoy being alive. Stela was pure presence: sun, grass, birds, just being near. While I was grinding mobile games, scrolling news, and procrastinating on the things that actually matter: talking to friends, spending time with kids and pets, fixing the house, stepping outside. I didn’t want another “hit your streak or you’re a failure” app. I wanted something that simply says, “You showed up 18 times this month. That’s good.”
Way Finder is that prototype:
- Aggregation, not streaks: “18 walks this month”, “12 study sessions”, “5 bathroom‑repair sessions”
- Aspirations → Habits → Activities: “Get stronger for hiking” → “Do pushups” → daily logs with values
- LLM‑assisted setup: “Remodel bathroom” or “Prep for certification” → suggested sub‑activities, metrics, and structure
- Dashboards and trends: totals, trends, and resistance breakdown instead of red Xs and broken streaks
- Projects and progress photos: phases, time tracking, and visual progress (with a clear path to shine on mobile later)
Under the hood it’s a Next.js 16 + TypeScript + IndexedDB (Dexie) event‑sourced prototype, deployed to Azure Static Web Apps with a custom domain at pushok.life, using GPT‑5‑nano to help define and configure goals. It’s not production‑grade; it’s a thoughtful, working prototype that validates the data model, the UX, and the “live better, not just grind harder” philosophy built in parallel with my day job.
Demo
- 🌐 Live Prototype: https://pushok.life
- 🎬 Demo Video: https://github.com/aleksey-cherenkov/pushok/blob/main/demo/demo-final.mp4
- 💻 Code: https://github.com/aleksey-cherenkov/pushok
My Experience with GitHub Copilot CLI
I built this prototype almost entirely through GitHub Copilot CLI while working a full‑time job, and the experience was both energizing and eye opening.
Copilot CLI made it possible to build in short bursts parallel to work without the usual mental overhead. I could ask questions, fix bugs, and ship features without switching tools or hunting for tutorials. That alone made development feel fun again, closer to how it felt in high school when everything was experimentation and momentum.
At the same time, this project showed me the real dynamics of working with AI at scale. Copilot often generated different implementations for similar features, different UI patterns, and even different backend shapes for dashboards that should have been consistent. It also tended to repeat the same mistakes every time it scaffolded a new API route. Normally I’d catch all of that through a proper workflow: separate branches, PR reviews, careful diffing, and tightening instructions so Copilot reuses established patterns.
But for this prototype, I intentionally didn’t do that.
I wanted to see what would happen if I stayed mostly hands off, letting Copilot CLI drive the implementation while I focused on workflow, usability, and the overall shape of the system. When something broke, Copilot (and Playwright) were great at uncovering and fixing bugs quickly. When something drifted stylistically, I let it drift. This wasn’t a production build, it was an experiment in momentum.
And that experiment worked. I shipped a complete prototype, learned a ton about how Copilot behaves across repeated patterns, and proved that for one‑off tools, helpers, or prototypes like this, the CLI can take you surprisingly far. For production work, I’d absolutely return to my usual discipline, reviewing everything, enforcing consistency, and guiding Copilot with tighter context. But for this challenge, the “vibe mode” approach was exactly what I needed to break Resistance and ship something real.
Development became joyful again. That’s the part that matters most.
It’s a fully working prototype—you can use it locally or at https://pushok.life, and all data stays in your browser. Comments, suggestions, and mobile ideas are very welcome. There are similar productivity apps on mobile, but most rely on monthly subscriptions. Now I understand why: once you go mobile, you need external storage, sync, and compute, and none of that is free. I’m thinking about how to keep costs minimal while still making the app accessible to more people. If you have ideas for low‑cost hosting, sync, or mobile architecture, I’d love to hear them.
Top comments (0)