Why don't I see the Push button in my AI chat?

Three usual reasons: the content isn't an HTML page (no button on Python scripts), the extension just updated (refresh the tab), or the AI site changed its interface and pushd needs an update. Meanwhile, the popup's drop zone deploys anything.

My deploy failed with a token error - what now?

The token was revoked, expired or mis-pasted. Create a fresh one from the provider's token page (pushd links to it) and reconnect - takes a minute.

Where can I see details about my last failed deploy?

Dashboard → About shows the last deploy error with provider, message and time. It reads 'None - all deploys healthy' when everything's fine.

Docs

Troubleshooting

Organized by symptom.

The Push button

No button at all. In order of likelihood:

The content isn’t an HTML page. pushd only offers Push on deployable HTML - Claude artifacts that are code/scripts/documents, or non-HTML code blocks elsewhere, don’t get one.
The extension updated in the background. Chrome leaves old tabs disconnected after an update - refresh the tab.
The AI site changed its interface. These sites redesign often; when they move things, the button can vanish until we ship an update.
Unsupported site. Only claude.ai, chatgpt.com, gemini.google.com and grok.com have buttons today. DeepSeek and friends are on the roadmap; their pages deploy fine via the drop zone.

⟳ Refresh page (tooltip: “pushd was updated - refresh this page to reconnect”). Exactly that - the extension updated while the tab was open. Refresh.

⚠ No HTML. The capture came back empty - usually the page hadn’t finished rendering, or a clipboard hiccup on Claude or Gemini. Let the page finish generating, click again.

⚠ Failed. Something unexpected in the capture. Refresh and retry.

Whatever the cause, you can always copy-paste your HTML into the popup, or download the file and use the drop zone.

Connecting a provider

The token field checks live, so most problems surface immediately as ✗:

Token looks wrong - double-check it in your [provider] account. Invalid, expired, revoked or mis-pasted (partial copies are common). Create a fresh token from the provider’s page - pushd’s Get token ↗ goes straight there - and paste again.
Token looks wrong - make sure it has “Workers Scripts: Edit” permission. Cloudflare tokens must be created from the Edit Cloudflare Workers template. A “read-only” or custom token without that permission fails. Guide.
…Disconnect [provider] first, or upgrade to Pro…“ Not an error - this is the free plan limit. Swap providers any time (disconnect, connect), or Pro connects everything at once.
Couldn’t reach your cPanel - check the address / cPanel rejected login - check username and API token. All covered on the cPanel page.

Deploys that fail

Connect a provider first - the deploy has nowhere to go; pick and connect a provider above the deploy button.
Re-connect [provider] - account ID missing - stale connection state (usually after a very old install updated). Disconnect and reconnect the provider; the token can be reused.
That site is too large to deploy - the site is too big. Usual culprit is video or huge images; try the Optimize pass (Pro) or slim the files.
Too many requests / rate limit hit Rate limits pass on their own after a certain period of time, you need to wait it out.
Your account has no workers.dev subdomain yet One-time Cloudflare step - how to claim one.
You already have a GitHub repository called “{name}” that pushd didn’t create - pick a different site name so we don’t touch it. A safety guard, not a failure - rename, or tick “Overwrite if this name already exists” if you mean it.
GitHub Pages could not build the site. Rare for plain HTML - re-push, then check the repo’s Actions tab on github.com.
Tiiny rejected the API key Tiiny’s API is paid-plan-only; email mode is the free path.
[provider] could not process the upload. Provider-side hiccup or malformed files - re-push, verify the files open locally.
That deploy didn’t finish The browser interrupted the background deploy (sleep, shutdown, extension reload) - push again.
Something went wrong - reload the extension and try again - the extension didn’t respond. chrome://extensions → reload pushd, retry.

Dashboard → About shows your last deploy error - provider, exact message, HTTP status and time.

The site deployed, but looks wrong

Showing an old version. Likely browser cache. Hard-refresh (Ctrl+Shift+R / Cmd+Shift+R) or try a private window. Providers with build steps can also take a minute to rebuild after an update.
404 on a multi-file site. There’s no index.html at the top level - that file is the homepage. Rename your main page and re-push.
Site gone. Tiiny email-mode sites expire after ~15 minutes by design. A Netlify site may have reached the free plan’s traffic cap - sites pause until the month resets.

History and rollback

No saved copy for this deploy. Redeploy needs the local snapshot, which is skipped for deploys over ~5MB (and gone after Clear history / data and uninstallation). Deploy that version from its original files instead.
Redeploy button missing on a row - same reason: no snapshot for that deploy.

Pro

Paid but Pro isn’t showing: Dashboard → Plan → “Re-check status”. On a new machine: “Already subscribed? Sign in to restore” with the same Google account. More.

Still stuck?

Email support@hypercollider.co.uk - include what Dashboard → About shows for the last error, and roughly what you were pushing (one file? folder? which provider?). That’s usually enough to solve it in one reply.

Add pushd to Chrome - it's free All docs