Monday, 29 June 2026

I Built a Daily Task Tracker Using AI in 30 Seconds — No Framework, No npm, Just One HTML File

As someone who's spent 14 years in DevOps — writing Terraform, managing Kubernetes clusters, debugging at 2 AM — you'd think I'd have my personal productivity figured out.

I don't.

Not the big things. The big things have tickets, sprints, alerts. It's the small daily tasks that slip through. Write a blog post. Apply to a job. 20 minutes of something personal. Gym.

Each one takes less than 10 minutes to start. But between terminal windows, Slack pings, and context switching — they disappear.

So I built a tool. And how I built it is worth a post.


The Build: 30 Seconds, Zero Dependencies

I opened Claude AI, described the problem in plain English, and got back a fully working dark-theme task tracker — single HTML file, no framework, no build step, no npm install, no internet required after load.

Features that came out of the box:

  • Color-coded categories — Blog, Jobs, Music, Gym, Other
  • Live progress bar with done/total count
  • Auto-resets daily (localStorage keyed to date)
  • Keyboard shortcut — Enter to add tasks
  • Toast notifications on every action
  • Delete and clear-done in one click

Open it in Firefox, Chromium, whatever you have. Drop it in ~/Desktop/tasks.html and it just works.

For Linux users especially — no Electron app eating 400MB of RAM. No account. No cloud sync. Just a file.


The Prompt That Built It

For the technically curious, here's roughly what I said to Claude:

"Build me a dark theme HTML page to add/remove and tick daily tasks. Categories: blog, jobs, music, gym. Progress bar. Saves between sessions. Auto-resets daily."

That's it. One paragraph. The output was production-ready HTML with scoped CSS and vanilla JS — clean enough to read and modify yourself.

If you want to customise categories, change colours, or add a Pomodoro timer — the code is readable. No minification, no obfuscation.


What This Means for DevOps and Tech Folks

We spend our days automating things for other people's systems.

AI is now letting us automate things for ourselves — with the same speed we'd spin up a container or write a bash alias, but for interfaces, tools, and workflows we actually use.

The skill that matters now isn't syntax. It's problem articulation. Knowing exactly what you need, described precisely. That's a skill every experienced engineer already has — we write runbooks and architecture docs for a living.

Point that same clarity at AI, and you get working tools in minutes.


Download and Use It

Save the HTML file, open in browser, done. No install. Works on any OS.

If you want the file or want me to add features — drop a comment below.




What would you build if execution time was near zero?

That's the real question this year.


Mahendra Shrivastava — Staff DevOps Engineer, builder of small useful things.

Also writing at kalyugrishi.com on AI, consciousness, and technology's deeper implications.


AI, HTML, Productivity, DevOps, Claude AI, Linux, No Framework, Build In Public

No comments:

Post a Comment