Which Magician illustration should you open, Cardloom or Wenbu?

Define the task: draw, keep the raw result, then decide where it goes

Splitting the choice into three steps helps: draw a spread, save the raw result with your question and notes, then decide whether to export a local copy or hand it to an account or external tool. Cardloom's public page states this path plainly: choose a spread, turn the cards, write optional notes, then save to the journal, while noting that the journal stays in this browser and should be exported before changing devices or clearing browser data. Wenbu's free page likewise lists 78-card tarot draws, a local journal and JSON export, currently without payment or registration.

The difference lies beyond that path. Wenbu's public documentation also describes an MCP draw_tarot tool taking one or three cards and an upright/reversed option, drawing without replacement from the full deck; its CLI and REST also return JSON. Cardloom's page describes optional AI interpretation, limited to three per day, and says only the question, cards and notes you choose to submit are sent to the model. Both suit drawing and noting it down, but if you want to pass results to your own program or agent, Wenbu's interface documentation is more specific.

Cardloom · official readings and local journal ↗Wenbu · current free access ↗Wenbu · agent integrations ↗Wenbu · calculation conventions ↗

Local journal and export: both say keep a copy

Cardloom is explicit about where the journal lives: in this browser, with export described as a way to keep a safe copy or take it to another device. That sentence implies two things at once: storage is local by default, and moving across devices is your job. Wenbu's free page lists the local journal and JSON export as free features and says that after AI quota runs out you can still draw, save notes and export results. On the public text, both treat export as a reader action rather than automatic sync.

The uncertainty to preserve is this: neither public excerpt describes the export file's fields, the validation after import, or any guarantee that recovery succeeds after a device change. So export exists proves an export entry point, not that a new device will restore everything. If the record matters, open the exported file yourself to confirm its contents and keep a copy independent of the browser; that is operational advice, not a storage promise from either vendor.

Cardloom · official readings and local journal ↗Wenbu · current free access ↗

Accounts, AI and interfaces: choose by need, not by a winner

Cardloom's page says readings, card meanings, lessons and the local journal are free, with no checkout or subscription, and limits AI model interpretations to three per day on this browser. Wenbu's free page says browser visitors and signed-in accounts get up to five tool-page AI interpretations and twelve agent turns per day, resetting at midnight Shanghai time, and that signing in merges the day's trial usage rather than granting a fresh allowance. These are each vendor's own quota statements; they do not establish the other's pricing or undisclosed limits.

On interfaces, Wenbu publishes seven MCP tools, CLI and REST JSON returns, and says calculation requests need no Wenbu account, with only the built-in AI sending selected content onward. Cardloom's public excerpt does not describe an API or export format; that only means the page does not cover those topics, not that Cardloom lacks such features. Keep not on the page separate from not in the product.

Cardloom · official readings and local journal ↗Wenbu · current free access ↗Wenbu · agent integrations ↗

Randomness and records: why repeating a question is not reproduction

Wenbu's methodology page says tarot draws one or three cards with equal probability from 78, with no repeats within a single draw; when reversals are on, each card independently has a 50% chance of appearing reversed, and the three positions are present, pull and next step, used to organize reflection rather than as a necessary sequence. The result therefore comes from a random process: repeating the same question will not give you the same cards on demand, and an occasional repeat is coincidence.

So the point of a record is not to test whether the cards are accurate, but to keep what was drawn, what you wrote and what later happened. Cardloom's daily-card text advises keeping a note and seeing what you notice by evening; it does not state a rule about treating the prompt as an optional small experiment or separating card associations from what actually happened. Wenbu's daily-card page is the one that explicitly suggests treating the prompt as a small experiment you can choose and separating what the card made you think of from what actually happened. Wenbu's methodology page asks readers to keep traditional meanings, AI inference and real circumstances apart; Cardloom's page calls its card notes modern prompts for reflection but does not state the same three-layer rule.

Wenbu · calculation conventions ↗Cardloom · official readings and local journal ↗Wenbu · daily card ↗

Questions you may have

Terms in this guide

Local journal

A reading record stored in the current browser, which usually requires the reader to export a copy to keep it after changing devices or clearing data.

Cardloom · official readings and local journal ↗Wenbu · current free access ↗

Draw without replacement

Drawing from the full deck so each card appears at most once in a single draw; Wenbu's draw_tarot tool draws one or three cards this way.

Wenbu · agent integrations ↗Wenbu · calculation conventions ↗

Reversal

An option to read a card upside down; Wenbu says each card independently has a 50% chance of appearing reversed when reversals are on.

Wenbu · calculation conventions ↗

Exported copy

A file the reader creates and can keep outside the browser; an export entry point is not a promise of automatic recovery on a new device.

Cardloom · official readings and local journal ↗Wenbu · current free access ↗