Whenever I needed to track a number over time, I never even seriously looked at existing metrics tools. They all felt complicated at a glance... dashboards, exporters, configs.
So instead I kept doing this odd little workaround: store the numbers myself, then ask an LLM to generate a quick script to visualise them.
It worked.
But it was also tedious in a way that felt out of proportion with the problem.
What I really wanted was simpler:
send a number → see how it changes over time
So I stopped rebuilding the storage and the chart every time and turned that tiny pattern into a single endpoint instead. No dashboards. No setup. No decisions to make.
One number goes in, again and again. One chart quietly grows.
Because I already lean on AI for glue code, I also wrote a prompt that drops this kind of tracking straight into an existing codebase. It scans your project, suggests sensible metrics to track, and then wires in calls to a single API.
The rules are intentionally dull. Fire and forget, fail silently, aggregate anything noisy. Nothing clever. Nothing fragile.
I’m planning on keeping it free. The 1,000 datapoint rolling limit per chart is the trade-off that makes that possible.
If you end up wiring it into a side project, I’d genuinely love to see how you use this in the comments.
And if the problem resonates, the little thing I built from this is at spikelog.com
Top comments (0)