Your best chart is the one nobody saw

Sharing work out of a Jupyter notebook is easy enough. Right-click a chart, copy, paste into Slack. Done. Two seconds.

But that chart is now a frozen image. Update your analysis tomorrow and the thing you pasted is wrong – and nobody knows it's wrong, because it still looks like a chart. If you also dropped it into a slide deck, that's two copies that are now silently stale. Update the deck because that's the high-stakes one, and now Slack has one version and the presentation has another. Your team is looking at two different pictures of the same thing and nobody realizes it.

That's the quiet cost. It doesn't look like a problem because nothing breaks. The old chart doesn't throw an error. It just sits there, confidently showing yesterday's numbers to whoever opens that Slack thread next week.

Then there's the loud cost. Presentation decks. For every chart that goes into a deck, you're pasting, resizing, nudging into position, fixing the aspect ratio. One chart, a few minutes. Across a full presentation, an afternoon. And when someone asks for a change – or you spot an error, or you think of a better way to present something – you weigh the improvement against the cost of redoing all that manual work. You end up talking yourself out of doing your best work because the logistics aren't worth it. That's a bad place to be.

And then there's the cost you never see. A lot of good work dies in notebooks and no one knows to miss it. The overhead of getting it in front of someone just wasn't worth the effort. The minutes you spend on screenshots and slide formatting are annoying, sure. The findings that never get shared at all – that's the real cost.

The notebook-alternative platforms – Deepnote, Observable, Hex – solve some of this. But they solve it by asking you to leave Jupyter. Well, I don't want to leave Jupyter. I have an environment I've spent years configuring, a coding agent wired into my local tools and context, and a workflow that works. Paying for a cloud platform just to solve a sharing problem is a bad trade.

So I built Skua.¹ The entire interface is one function call:

import skua
skua.snap(fig, title="Q3 Revenue Trends")

That prints a URL. Send it to whoever needs it. They see the chart in their browser – if it's a Plotly figure, it stays fully interactive. If it's a DataFrame, it renders as a searchable, sortable table. No screenshot, no paste, no resize.

Re-run the cell and the content behind that URL updates. Same link, current data. The Slack link and the one in your notes and the one you emailed last Tuesday – all the same, all current.

No configuration, no API key, no signup. Install it, call snap, get a URL. If you verify your email, your findings stick around longer and get more privacy controls. If you're wondering where your data goes – fair. Here's how Skua handles it.

I have a lot of ideas about how to fix the rest of this. Some will work, some won't. But I know what I'm solving for. The work that data scientists and analysts do gets its value from being seen – by your manager, your team, the decision makers. Those people are never going to open a notebook. They need a link they can click. And the person who did the work shouldn't have to jump through hoops to give them one.

When sharing is tedious, it happens less. When it's manual, it goes stale. I want sharing to be easy enough that it happens often, and reliable enough that it stays accurate. I'm already thinking about what it would take to update every chart in a slide deck with one click. The friction between "I figured something out" and "the right people can see it" should be as close to zero as I can get it.

If any of this sounds like your Tuesday, pip install getskua. I know adding a new tool to your workflow isn't a small ask – there are practical considerations around security, trust, and whether the thing even does what it says. I wouldn't adopt something like this without kicking the tires first. So kick the tires. If something doesn't work, or could work better, I want to hear about it. You can reach me at kal@skua.dev.

I'll be writing more here about notebook workflows, the gap between doing good work and getting it seen, and the tools and patterns that actually help. Thanks for reading.

– Kal

---

1. A skua is a seabird that will chase down birds three times its size and dive-bomb anything that gets near its nest. Not beautiful, not well-known, but scrappy and so good at what it does that it's earned a bit of a reputation. I think they're pretty badass.