Transform logs

A record of every transform: what ran, what it built, what failed, and how much AI it used.

The list

Each row is one transform run — everything you started with a single action, such as transforming a selection of Sources, or one Source on its own. A row shows what it processed, what started it (Manual, Onboarding, Webhook, After sync, AI assistant, or System), when it started, how long it took, and how many items were created, updated, completed, skipped, or failed.

Status is Running, Completed, Completed with errors, or Failed. Completed with errors means some items failed while the rest came through. Items still waiting show as Queued or Processing. AI tokens is how much AI the run used, when it used any. See AI token spend.

A run appears as soon as you start it. The page refreshes on its own while anything is running. Needs attention shows only runs with failures, and Running shows runs in progress. You can also filter by what started the run.

Open a run

Select a row to see the items, a count of each result, and a summary of what the run did: blocks created, updated, left as they were, removed, or published, blocks kept because you had edited them, steps that were turned off, and images copied or not.

Each item shows the Source, the Cube it built, its status, and a message. Open the Cube to see the result. A product transformed in several languages is one item.

A Source with little in it, such as a product with no images or description, gives a transform little to build from. The message says what went wrong.

What it did

Open an item to see what changed:

• Cube and Blocks — whether the cube and its generated blocks were created, updated, left as they were, or removed.
• Kept as-is — blocks left alone because you had edited them. See Keeping your content up to date.
• Process steps — which steps were on or off. A block you expected and did not get is often a step that was off.
• Languages — whether a language used its own translation, or the default language when no translation has been synced yet.
• Media — what happened to each image. See Media files.

What a second run does to blocks is in Running a transform again.

Media files

Rubikc keeps its own copy of each image a cube needs. Each file is one of:

• Uploaded — a new or changed file was copied.
• Already stored — the copy Rubikc has was still current.
• Skipped — left alone on purpose. Videos are never copied, and a link that is not an image is not copied either. This is not an error.
• Failed — the file could not be copied, and the reason is shown. It is left out of the cube. Once the image is available in the source system, run the transform again.

Retry

A run with failures has Retry failed. It runs only the failed items again. A failed item has its own Retry.

AI token spend

Some transform options use AI to write text, such as a card’s subtitle or a short description. The page shows how many tokens those calls used — tokens, not a price — on the run, on each item, and as a total for the Space over the last 30 days. A transform with no AI options turned on shows no tokens.

How long it is kept

Item details are kept for 14 days. The run, with its counts and token totals, is kept for 90 days.

After a failure

Fix the data in the source system, sync again, and retry the failed items or run the transform again. Editing the cube by hand also works for a one-off.

For sync problems, see Sync logs.