// changelog
Every entry is written when the change ships, by the people who shipped it, and it is the same list the product shows inside your dashboard.
// the record
Connect a Reddit ad account and the Brain can pause, resume and re-budget campaigns there on the same rails it already uses for Google — the same approval model, the same spend ceilings, the same decision log. Reporting reads back from Reddit rather than from what we asked for, so a change that did not take shows as a change that did not take.
Linked products could be created and never listed. The page now shows every link, what each one shares, a switch per shared field, and an Unlink on both the group and the single product. Sharing cost data is separate from mirroring a price or an archive, and each is its own control.
It was mounted on one of five app shells, so Launch, Workspace, Agency and the welcome flow had no way to report anything at all. Every surface with a signed-in reader now carries it, and notes are filed against the page you were standing on.
The agent's credit spend now renders on its own page beside its mode, because either number alone reads as fine: a spend is unremarkable on a desk that answers customers, and "Dry run" is unremarkable until you see what it costs. The four modes are named individually instead of collapsing into "Approval".
A new Links chip appears between Export and Suppliers on the Catalog toolbar, so you can check whether a partner store carries the same products without leaving the page you are already on. The chip is hidden when your workspace has only one connected store.
Products across your connected stores are now scanned weekly for matches using shared SKUs, supplier images, and product similarity — so you no longer need to trigger linking manually. A review queue shows you candidates with the evidence behind each suggestion before anything is confirmed. Confirming a link shares data between stores but never moves a price.
The ticket link panel now shows whether a customer has viewed, replied through, or never received their portal link — so agents know at a glance whether a reply has been seen. If a link was forwarded to the wrong person, agents can now revoke it; the customer automatically gets a fresh working link on the next send while the old one stops working immediately. Previously there was no way to cut a portal link from inside the dashboard at all.
Customer-facing replies sent from your helpdesk now include a masthead, a preheader, an author block with the signer's name and photo, and a ticket reference number — matching what customers expect from a professional support email. If you haven't uploaded a photo for your support agent, their initials appear on a branded monogram instead, so every reply looks intentional from day one. Upload your agent's photo under CS → Settings → Ticket page to personalise the author block.
The support ticket portal was missing a favicon, so it appeared as a blank tab icon next to your shop's branded tabs — this is now fixed with a full set of icons including mobile home-screen support. The portal also gains the same security headers the main dashboard already had, including stricter cross-origin resource policies and HSTS across all subdomains.
The Tracking page now shows every enrichment signal Magistry sends to Google, with a separate indicator for whether you have enabled it and whether your store actually holds that data. Signals exclusive to Magistry — including cart contents, customer value band, and profit instead of revenue — show the specific reason no tag vendor can match them, so you can verify the claim yourself. Coverage figures only appear when they have been measured.
Automated reply emails no longer include a disclosure stating they were written by an AI assistant. The link to your support ticket is now a clearly styled button that renders correctly in Outlook and on mobile screens, replacing a small grey URL that was easy to miss. All three email sending routes now include the ticket button.
Magistry now sends Google Ads a customer value signal — LOW, MEDIUM, or HIGH — based on each buyer's real spend history with your store, so your campaigns can automatically bid more for the customers worth having. The signal is ranked within your store, not against an absolute threshold, so it works correctly even for newer stores. When there is not enough data to be confident, no signal is sent rather than a wrong one.
Google conversion events now include the buyer's browser type and IP address alongside the hashed contact details already being sent. Google uses these signals to match events to clicks or users it recognises, and drops events it cannot join — so more of your conversions will now be attributed correctly. Device details are withheld whenever a customer has explicitly denied consent, the same gate that protects all other personal data.
The Brain, Website, and Social sections now display a "Beta" badge in the navigation, so you can easily identify which features are in early access. These features also appear under Settings > Beta features alongside other Beta areas.
A new setting on your conversion goals page lets you decide whether Magistry reports refunds back to Google after a purchase is tracked. Off by default, so your conversion count stays comparable with tools like TrackBee that never send refund adjustments. Turning it on gives Google a net-of-returns figure, which can steer bidding away from buyers who return — but will show a lower number than competitors tracking gross revenue.
Your integrations and AI agents can now discover Magistry's API capabilities automatically via the standard /.well-known/mcp.json address. The versioning guarantees previously buried behind a broken link are now embedded directly and machine-readably in the contract document, so your CI can assert on the notice period as a number without parsing prose.
Magistry's REST API now has a generated TypeScript SDK with typed arguments for every operation, an llms.txt file so AI agents can discover the API, and a server.json MCP registry manifest. All three are derived from the same OpenAPI document that the API already serves, so they stay in sync automatically.
// direction
// stay current
// changelog list
// e-mails not sent yet · no tracking pixels
// see it run
Connect your store and choose the Brain's mode — in dry run every decision is logged with what it would have done, and nothing touches your store.
The Brain changes nothing until you choose a mode · Every action logged · Price, status and copy changes can be rolled back
We use cookies to run this site, to see which pages help, and to measure which ads bring people here. You choose per category; everything except the strictly necessary is off until you say otherwise. Cookie policy · Privacy
