I Built a Tool to Save Curators From Burnout

Alright, the title might be a bit clickbait-y :D

If we’ve ever done manual curation on Hive, really done it, consistently over time, we know how it goes. We open our favorite frontend like PeakD or Ecency, start scrolling, and within minutes the pattern becomes familiar. A few “just my thoughts” posts with barely any substance, a stream of reblogs, maybe a couple of questionable posts that raise eyebrows. Before long, we realize we haven’t found a single thing worth voting on.

So we close the tab, tell ourselves we’ll come back tomorrow, and move on. But tomorrow looks the same, and the day after that too. To me that could be a symptom of curator fatigue.

I’ve been there more times than I can count, and after sitting with that frustration for a while, I decided to build something to fix it: Hive Search. This tool is not new by the way, I’ve been using it in my manual curation work for @curie for some time. It started as a localhost-only tool, something just for me, but at some point it felt a bit silly not to share it.

At its core, this is what I’d call a curation cockpit. Basically a web app designed specifically for people who actively curate on Hive. We log in with Hive Keychain, set our filters, and get a clean, real-time feed of posts that actually match what we’re looking for that day. It pulls data directly from HAFsql, so everything we see is based on raw blockchain data, not some delayed or reshaped feed.

The part I personally find most useful is the preset system. The reality is, our standards as curators aren’t static. Some days we’re in discovery mode, actively looking for new voices, willing to give a 45-rep newcomer a chance if the content feels genuine. Other days we’re simply tired and want posts that almost curate themselves, those from higher reputation authors, longer form, proven engagement. So instead of constantly tweaking filters back and forth, we can just save those mindsets as presets. “Strict Mode,” “Forgiving Mode,” or whatever naming makes sense. Then it’s just one click, and the feed adapts instantly. It’s a small feature on paper, but in practice it removes a surprising amount of friction.

From there, everything else is about giving us control without slowing us down. We can dial in reputation ranges, filter by pending payout to surface under-rewarded posts (or avoid ones that are already discovered), set word count for long-form content, and even filter by image count for photo or art curation.

There’s also something I’ve been experimenting with called Engagement Quality Ratio, or EQR. It looks at an author’s comments over the last seven days and measures how many of them are actually substantive—defined here as more than 20 words. The idea is simple: distinguish between people who genuinely engage and those who just post and disappear. But it’s important to treat it as a signal, not a judgment. A great writer who rarely comments will score low, and someone who writes thoughtful but concise replies might get underrated. We can click the EQR label to inspect the comments ourselves, which matters more than the number.

Everything updates live as we adjust it. No apply button, no extra steps. We just move a slider or tweak a filter and the feed responds immediately, keeping us in the flow instead of breaking it.

There’s also a small badge next to each author called KE ratio. It’s calculated from current Hive Power and lifetime power-downs, showing how much someone has taken out versus what they still hold. A low KE suggests they’ve mostly kept their stake, while a higher one means they’ve extracted more over time. It’s not meant to judge anyone’s choices, just to add context when deciding where our vote goes. If we’re curious, we can click it and see the breakdown.

The rest of the experience is built around staying in that same flow. When we find something worth reading or sharing, we can open it directly on our frontend of choice: PeakD, Ecency, InLeo, or Hive.blog, without copying links around. We can also vote right from the preview panel using a slider from 1 to 100 percent, with Hive Keychain handling confirmation. The app checks if we’ve already voted, so we don’t accidentally double up. If we want to leave a comment, there’s a box for that too, using the same quick confirmation flow.

On the backend, it quietly syncs with the hivewatchers blacklist every day, so accounts flagged for spam or plagiarism are filtered out before they ever show up. A small indicator lets us know when that data is current, but otherwise it stays out of the way.

I tend to think that everyone on Hive is a curator in some sense, but this is especially useful for those of us doing it regularly, whether as part of a guild like Curie, OCD, or Curangel, leading a curation team, or simply getting tired of spending 30 minutes scrolling and ending up with nothing to vote on.

Right now, this is still an MVP. There are a lot of directions it could go, like analytics for guild leaders, deeper insights into curation patterns, detect AI generated content, maybe even AI-assisted discovery based on past voting behavior, but those can come later.

I certainly hope that you guys will give this tool a try.

One thing worth mentioning: the app is mobile-friendly and works well on smaller screens. That said, Hive Keychain as a browser extension doesn't play nicely with popular mobile browsers like Chrome or Brave, so logging in or voting from those won't work as expected. The workaround that works great is using the built-in browser inside the Hive Keychain mobile app, everything flows smoothly from there.

Feedback is always welcome.

Check it out: https://search.hive-roshambo.com