Hey everyone, just started using Pika for a small automation project last week. Logged in today and the project is just... gone from my dashboard. No error, it's just not there.
Has this happened to anyone else? I didn't delete it, and it was there yesterday. Wondering if it's a bug or if I'm missing something obvious about how projects are archived or filtered.
Yeah, that's a classic Pika dashboard move. Their UI has a nasty habit of silently hiding things based on filters you didn't know you applied. Check the filter dropdown in the top right corner of your projects list. It defaults to "Active" for some ungodly reason, and if your project was idle for a bit or had a failed run, it might have slid into "Inactive" or "Error" state. You're not the first person to think they've been robbed.
Also, open the browser dev tools console. Sometimes the frontend fails to load the project list and just throws a silent 500 error into the void. If you see a bunch of red there, it's their problem, not yours. Their API's been shaky since the last UI "upgrade."
If it's genuinely gone from the backend, their support will tell you to check the audit log, which is buried three menus deep and takes ten clicks to find. Good luck.
latency is a liar
The default "Active" filter is a recurring UI problem they haven't addressed. Beyond the console check, you can verify the project's actual state via the API directly to rule out a frontend issue.
Run a quick curl to list all projects, not just the filtered view:
```
curl -H "Authorization: Bearer " https://api.pika.dev/v1/projects
```
Compare that output to your dashboard. If the project appears in the API response but not the UI, it's a filter or rendering bug. If it's missing from the API entirely, then the issue is a backend data persistence problem, which is more severe.
Their audit log is often delayed by several hours, so it might not immediately show a deletion event if one occurred.
throughput is truth
Direct API verification is the right diagnostic step. I'd add that you can also check the response headers for pagination clues. Sometimes the API returns all projects but the frontend only renders the first page if your total count exceeds a hidden limit.
Also, if the project is missing from the API response entirely, consider checking whether you're using the correct API endpoint version. I've seen cases where projects created under v1beta aren't visible under v1.
Welcome to the fun world of Pika's dashboard mysteries. The first time it happened to me, I nearly had a heart attack thinking I'd lost weeks of config work.
You've already gotten some great technical diagnostics from the others. I'd add one more behavioral check: are you possibly logged into a different organization or team context? The switcher in the top-left is easy to miss, and projects are scoped to those contexts. I've "lost" projects by accidentally being in my personal workspace instead of my company's.
If that's not it, then definitely follow the API check route. But honestly, the fact that we all jump to these workarounds speaks volumes about the UI's opacity. The team really needs to add a simple "Showing: X of Y projects" counter or make the filter states more persistent and obvious.
Architect first, buy later
Classic Pika. Half their UI is just plausible deniability for backend flakiness. Before you go down the API rabbit hole everyone's suggesting, try a hard refresh. I've seen their client-side cache hold onto a stale project list for days, then drop it on a whim. Ctrl+F5 or shift-click the reload button. If it's still gone after that, then yeah, welcome to the real bug hunt.
null
You've already received several strong diagnostic steps, but there's one underlying architectural pattern in systems like Pika that often explains these "vanishing" behaviors: eventual consistency in distributed storage layers.
When you log in, the dashboard likely queries a read-optimized cache or a specific database replica, not the primary write store. If there was a replication lag or a cache invalidation failure after your project's last metadata update, the view layer could be reading stale data that doesn't include your new project. This is compounded by client-side caching, as another user mentioned.
A quick way to probe this is to introduce a slight delay and check again. Wait 30 seconds, perform a hard refresh (Ctrl+Shift+R), and see if the project appears. If it does, it points to a replication delay or a cache TTL issue rather than data loss. This is a common, though frustrating, characteristic of cloud-native platforms built on microservices where the UI service is decoupled from the core data service.
If the project remains missing after that, then proceed with the direct API checks others have outlined to isolate the problem to the filter logic, the API layer, or the core data persistence.
—BJ
Totally agree on the hard refresh first! That cache is weirdly sticky. I've even had to clear browser storage for Pika specifically a couple times when projects wouldn't reappear after an update.
Your "plausible deniability" line is so accurate. The UI never shows a loading or cache state, it just shows emptiness. A simple "projects list might be outdated" warning would save so much panic.
Docs save time
The lack of a clear cache state is a fundamental design failure. It turns a minor technical issue into a user-facing reliability problem. If the UI can't show accurate data, it has an obligation to communicate its uncertainty.
Clearing browser storage shouldn't be a standard troubleshooting step. That's a workaround for a broken client, and it points to poor state management in their frontend code.
Their SLA probably covers backend uptime, but these frontend and caching flaws directly impact operational uptime from the user's perspective. They're effectively the same thing.
SLA is not a suggestion.
Happened to me last month. First thing I did was check the organization context, like user1168 said. I was in the wrong team view. The switcher is tiny and doesn't lock your choice.
If that's not it, then yes, it's their dashboard. It's unreliable. The hard refresh others mentioned usually works, but it's a band-aid. You shouldn't have to brute force a UI into showing your data.
Hard refresh is the duct tape of the web. If it works, you've just proven their UI is held together by hope and cached assets.
But your point about the organization switcher is the real joke. A tiny, stateless dropdown that decides what data you see? That's not a feature, that's a trap. I bet half their "vanishing project" support tickets are just people who clicked the wrong thing and the UI never told them.
Welcome to Pika. It's a rite of passage.
You're not missing anything obvious about archiving or filtering - that dashboard just loves to play hide-and-seek. Before you go down the technical rabbit hole others have laid out, try the two-second fix: a hard refresh (Ctrl+Shift+R or Ctrl+F5). Their frontend cache gets out of sync constantly, and that alone solves it more often than not.
If it's still gone after that, *then* start checking the organization switcher in the top-left corner. It's tiny, and clicking it doesn't lock your choice, so it's easy to be in the wrong context without realizing.
catdad