Changes to Platform

A deployment's page now says which revision of your company's deployment template, or of Corefinity's default, each stage was built from, and which stages that environment sets for itself.

Use the Feedback button at the bottom right of any page, or press Shift+F, to report a bug, suggest a feature or ask for a new platform. You can add a screenshot of the page and mark it up; passwords and other secrets on screen are hidden automatically.

Set your build, database and after-deploy commands once per platform; every environment follows them and can change any stage for itself.

Every CoreStack site now deploys in three stages, build, database and post-deploy, which you edit on the Deployments tab, each line with its own step and time limit. Existing pipelines are converted automatically, a group of sites at a time, and deploy exactly as before; credentials written into commands are moved into encrypted variables.

Pick the Node.js version and the global npm packages each environment runs, on its Runtime tab. Changes apply straight away, and deployments build with the same version and packages.

AI agents such as Claude Code, Codex and ChatGPT can now work on your CoreCluster and CoreStack sites through the control panel's MCP server. Create a token or connect an app under Company settings › MCP & API Access; every call is recorded there.

Standard date and time fields in the Control Panel now follow UK time, automatically switching between BST and GMT.

A new empty site or environment now shows a page confirming it is set up, with its services listed, until you deploy your code or upload files over SSH/SFTP.

The Create window describes CoreStack Enterprise as enterprise multi zone managed hosting for high traffic websites.

On Billing, Payment methods, Add card is now in the Cards box in the same style as Set up Direct Debit, and both buttons sit clear of the text above them.

The Billing overview and Usage pages no longer show the note that prices are per hour (the monthly price divided by the hours in the month).

The invoice PDF and the invoice email no longer carry the line about prices being per hour, matching the Billing pages.

When you order while signed in, the Verify step no longer shows a paragraph about your email and two-factor sign-in. The box with your name and the set-up below it say what to do.

On the Review step, an included staging environment is now listed by its name and size, with Included shown once beside it.

While we finish your order, the page checks every few seconds and shows a spinner only while it checks. It says when your site is being retried, or when the region has no room yet.

The email telling you your site takes a little longer to set up no longer says that nothing is charged until then.

The main button in every email from the control panel is now Corefinity blue.

Add a card now looks like the rest of the control panel. Default and Backup each say what they mean, and anything that goes wrong says what to do next.

With a default and a backup card already saved, Add card explains the limit instead of opening a dialog that can't finish. Menu actions that aren't available yet say why.

The order pages have more room between sections, fields and summary lines, and a long email address in Your order wraps cleanly. When something needs fixing, the page now takes you straight to the message or the field. The Billing step no longer repeats that UK VAT is added: the order summary shows it.

The deployment output is light in the light theme and keeps its dark terminal in the dark theme, on every deployment page. Masked values show as chips, and one line under the log explains how masking works.

Every sign-in screen now shares one cleaner design: the control panel login, signing in during an order, password reset, two-factor codes and invitation sign-up. The Google and GitHub buttons carry their official logos, and how you sign in is unchanged.

If you change a pipeline while a deployment is waiting to be run, the deployment now uses your change when you press Run Now. Before, it ran the commands from when the deployment was created, and only a retry picked up the change.

On the Managed Websites list, CoreStack Enterprise environments showed as Production rather than their own names. Each environment now shows its name again.

A payment method saved with Stripe Link now shows as Link with its masked email, not as a card with blank digits and an expiry of 00. The … menus open above the page without scrolling it, the sign-in page is blue to the bottom of the window, and a request that gets no answer shows a retry message instead of a script error.

A buyer who was signed in before ordering and confirmed their email with the emailed code was then told their email was already verified, with no code to enter. The Verify step now says only that the email is verified, which is true either way.

A CoreStack order is no longer taken when the region has no room for the site and its staging environment. You are asked to choose another region or contact us, and an order already waiting is set up as soon as there is room.

If your bank confirmed the card but the save didn't reach us, Try again now finishes saving it instead of failing.

Problems on your standalone CoreStack sites now appear in Needs attention, as they do for the sites on your CoreClusters: on the Managed Clusters dashboard, even when you have no CoreCluster, and on each site's own page.

When your company already has an account with us, the Billing step of a new order now fills in your billing address and VAT number and offers the card or Direct Debit your company already pays with, so there is nothing new to enter. Companies that already keep two cards are no longer stopped with an error.

A newly created site could refuse a copy, a branch environment or a files-only or database-only restore for up to 15 minutes after it was ready, saying its storage was handled by our team. Its storage is now checked as soon as it is set up, and until that check is done these requests say the storage is still being set up.

When a deployment command prints the value of one of your deployment variables, the log now shows it masked, on every deployment. Very short values and simple flags such as true or yes stay readable.

Creating a site or environment on a cluster whose configuration had not changed for an hour was refused with a message that the cluster's inventory was out of date. The regular check now keeps the inventory current, so the site is created straight away.

Deleting a site or environment whose creation failed no longer fails with 'could not be read for its final snapshot'. It is removed without a final snapshot, as it never held any data.

If Google Cloud runs out of the usual Arm machines (C4A) while a package change creates your cluster's new node pool, the new pool is now created on the next Arm machine type (N4A) instead of the change failing. Your sites move only once the new nodes are ready, as before.

When a deployment failed part way through, the site's scheduled tasks (crons) could stay switched off until someone retried or stopped it. The Control Panel now switches them back on as soon as a deployment fails.

Staging-only site pages now use the visible environment for page-scoped URLs, authentication, settings, and activity while retaining the hidden root as the group reference.

When a cluster cannot take a new site or environment yet, the request is refused at once with a clear message and nothing is created, instead of failing part-way with a platform error. Our team is notified automatically.

Restoring a snapshot of a large site no longer stops after 40 minutes of copying: the copy now gets time in proportion to the snapshot size, up to six hours.

The list of deployments on an environment, and the full deployments list, no longer need scrolling sideways to reach a deployment's status, dates and buttons, whichever way you opened the page. On smaller screens the Tasks and Duration columns step aside to make room, and both are still on the deployment's own page.

A deployment stopped part way through could not be retried: it sat as queued and never started on its own. Pressing Retry once now starts the replacement straight away, and where a deployment does have to wait, the page says what it is waiting for.

The clock icon on tables that refresh on their own, such as Actions, Tickets and an environment's Deployments tab, is now a labelled Auto-refresh on / off switch. Switching it off keeps every table still as you move between pages, reload or open a new tab, until you switch it back on or log out.

You now stay signed in for up to 12 hours of inactivity instead of 2. Two-factor authentication is still required every time you sign in.

Website names on the deployments screen are now links through to the website's own record, the same as website names elsewhere in the panel.

Breadcrumbs now follow a record's real ownership chain — company, then website, then environment — with every step a link, instead of always pointing back at the listing page.

The registration screens no longer clear the whole form when something fails validation. Each error is shown against the field it belongs to, including the individual password requirements.

The View icon on a deployment's tasks table now sits at the left of each row instead of off to the right, so it can be reached without scrolling sideways.

An environment's Deployments tab now has a single Deploy button in place of Create Deployment and Deploy Commit. Its page deploys a commit from one of the environment's Git repositories, or an uploaded release archive, and opens on the commit where Git is set up. Once you press Deploy, you land on the new deployment's page to follow its progress.

The Slack channel dropdowns on an environment's Deployments tab opened with an empty list. They show your workspace's channels again, so you can choose where deployment alerts go.

Tables that had lost their styling after the recent control panel upgrade, showing cramped cells with no heading row or dividers, are styled again to match the panel's other lists.

Most of the output on a deployment used to appear in red, even when the deployment had succeeded. The output now reads in a single colour, and red is kept for lines that report a genuine problem and for a step that actually failed.

Raise a Ticket on a failed deployment now opens the ticket form with the company, website, environment, department and subject filled in, and the description already written: which deployment failed, when, at which step and with which exit code. Add to it or write over it before you send.

Hovering a link used to change its colour so slightly that it was hard to tell anything was clickable, and on some screens, such as the ticket header and the dashboard's Websites and Jobs cards, links were not marked as links at all. Every link now underlines and deepens in colour when you point at it or reach it with the keyboard, and the ones that were unmarked are marked.

When a deployment fails at Checkout Repository, the task now says what went wrong - that the environment's repository has been changed or removed, or that the commit is not in the repository that was cloned - instead of only "Fetching repository failed!".

Images pasted into a ticket from Google Chat are saved with the ticket instead of linked back to where they came from, so they still display later on rather than turning into a broken image.

Opening a PDF or other file attached to a ticket no longer fails with a 403 error. Anyone who can see the ticket can open its files.

Agency users can close tickets for the companies they work with, which previously failed whatever their role.

Connecting a Bitbucket account no longer fails with a 500 error part way through authorisation.

The last step of company registration tells you when an email address is not a valid format, instead of accepting it and landing on a 500 error.

Tab bars across the panel sit against the left edge at their natural width instead of stretching across the screen, so the individual tabs can be told apart again. Long tab bars still scroll.

Retrying a previous deployment no longer fails before it reaches any of its tasks when there is no crontab backup to restore.

Uploading a release archive for a manual deployment no longer fails with the artifact being rejected, including archives that link into their own shared tree. Staff can also download a deployment's release archive.

Bullet points and numbered lists show their markers again in ticket replies, and the divider above the delete button no longer crowds the end of the message.

Profile > API Tokens generated the token but never displayed it, so there was no way to copy it. The pop-up now appears, and so do the confirmation pop-ups for resetting Google 2FA and for verifying your mobile number, which had stopped appearing in the same way.