Day 15 of 100 of building a tool everyday

I’m building 100 tools in 100 days for people around tech and recruitment, but it’s a loose idea. I want to build tools that I find fun or useful, or that help market my product ideas.

I read the DuckDuckGo founder’s book on marketing, Traction, many years ago. In the book, he talks about engineering as marketing. The idea is to build free tools, calculators, widgets, and microsites that are genuinely useful, then let them pull in traffic, links, awareness, and leads.

With things like Cursor, Claude, and Codex, the challenge isn’t really the code anymore. It’s coming up with ideas.

Personal software

We are now in the age of personal software. People fire up Claude and build something themselves in an hour or two that scratches their own itch, and I’ve applied that idea here.

I’m building urlcv.com, an AI-powered anti-ghosting recruitment platform, and as I go through marketing, sometimes ideas just come to me. For example, I built a meeting scheduler, a bit like Calendly, but without needing to log in.

Another tool I built was an SEO checker, where you can enter a page and audit it for common issues.

Both of these tools were built to fix a problem I had there and then. Each one took less than an hour or two to build.

I also built a desktop Mac app that allowed me to add speech bubbles to screencasts, because the Mac app Screen Studio that I use didn’t have that functionality.

One tool that started doing well on Reddit, in one of the UK subreddits before getting deleted, was a commute cost calculator. The mods pulled the plug.

The tool that a lot of people have enjoyed most is a resignation letter generator.

My process

I keep a Google Sheets list of ideas and rearrange them as I think of new ones or fall out of love with old ones.
I have a master prompt stored in Apple Notes which does a lot of the groundwork for me.
Once I start a new tool, I paste in the master prompt and then add as much detail as I want in either Claude Code or Cursor.
I test it and keep tweaking the tool with prompts until it gets to the level I want.
I have GitHub CLI installed locally, so it creates the repo for me in the URLCV organisation, creates the Composer package, and then pulls it into the main repo.

My prompt

We have a /tools section that loads tools dynamically via Composer packages. New tools should follow this exact pattern:

1. Prefer frontend-only (Alpine.js, mode() = 'frontend') unless server computation is genuinely needed
2. Create the package locally under packages/tool-{slug}/ with the standard structure: src/Laravel/{Name}Tool.php (implementing ToolInterface), {Name}ServiceProvider.php, resources/views/{slug}.blade.php, composer.json, and README.md
3. Register temporarily as a path repo in the main composer.json and add to config/tools.php, then run composer update urlcv/{slug} and test locally
4. Push the package to a new public repo under the urlcv GitHub org using gh repo create urlcv/{slug} --public ...
5. Switch the composer.json entry from path to vcs pointing at the GitHub URL
6. Run php artisan tools:sync and php artisan sitemap:generate
7. Commit and push to master to deploy (no PRs — single developer, merge direct) ask the user for approval first as the tool needs validated
8. Do not attribute Claude or Cursor in any commit messages

The tech stack

It’s built with Laravel, MySQL, and Alpine, and it’s very lightweight. The whole site runs between GitHub and Laravel Forge. Every bit of code written has been open-sourced and added to GitHub under URLCV.

Here’s what I’ve learned so far

Even if a tool starts getting traction on Reddit, mods may still delete it.

Funny tools do well on LinkedIn.

The hard part now is taste when building software. You can build a tool in an hour or two but will people use it, will be another story?

Has it translated into signups?

No. I believe this experiment may show that my effort might have been better spent elsewhere. Google seems to have a phobia of sending me traffic. It won’t list new pages.

I’m currently on day 16 of 100, and I’m keen to get ideas on what people would find valuable. Any tools you would like built?