Write down what you asked, how you understood it at the time, one action you can try, and a review date; add observations later. Leave out ID numbers, home addresses, medical records, passwords and other people's private details. Keep the original question and first interpretation intact, then append a dated paragraph instead of editing the old reading to match the outcome. Before saving, choose a location: guest browser records, signed-in account history and separate exports have different deletion and sync rules. An export may contain birth details and private questions, so read it before sharing.
Wenbu · reviewing a reading ↗Wenbu · current free access ↗Wenbu · agent integrations ↗
- Keep only reviewable context: question, first interpretation, action, date, new information, what did not fit and the next decision.
- Where you save changes what deletion and sync mean: browser records, account history and exported backups are separate.
- Preview exports and remove birth details, private questions or other people's information that the review does not need.
Journal save-and-export choice map
- 01Remove sensitive details first
Ask whether the entry contains identity numbers, addresses, medical records, accounts or another person's private information. Abstract it where possible and keep the decision structure.
- 02Write the first part that day
Add the date, original question, first interpretation, one executable action and a review date; do not write the later conclusion yet.
- 03Choose a save location
Guest records stay in the current browser; sign in for cross-device account history; clearing site data removes unsaved browser records.
- 04Preview before exporting
An export may contain birth details and private questions. Read it, remove what the review does not need, then decide where the backup goes.
- 05Give an Agent only relevant context
The web export previews chart, question and optional background. It does not upload automatically; the external host decides its own data handling.
- 06Append on review day
Add a dated paragraph with action, new information, what did not fit, alternative explanations and the next decision; export the old version before an important edit.
Wenbu · reviewing a reading ↗Wenbu · agent integrations ↗Wenbu · current free access ↗
Set the boundary before you write
A journal is useful because it preserves what you understood then and what you observed later, not because it stores a life file. Worth writing: the question you actually asked, the concern behind it, the wording that struck you, one small action and a review date. Not worth copying: identity numbers, exact addresses, financial account details, medical specifics, other people's names or private feelings. If the question depends on sensitive background, abstract it: “Should I join an eight-week course? I know the fee, not the workload” keeps the decision structure without naming an institution or a person.
Write the boundary for yourself too: this is cultural reflection and personal observation, not a diagnosis, legal opinion or investment basis. Replace “the cards said I would fail” with “my association was a worry that I would not keep going.” That lets a later review compare observable actions and feelings instead of treating a random draw as a forecast. Tarot and coin draws are documented random procedures; keep the returned result as it came. Repeating a question does not reproduce the same card or spread on demand, so record what was drawn rather than only the conclusion.
Wenbu · reviewing a reading ↗Wenbu · agent integrations ↗Wenbu · versioned public capabilities ↗
A fillable, privacy-conscious template
The template has two parts. Fill the first on the day of the reading: date, original question (abstracted if needed), first interpretation, one executable action and a review date. Fill the second on the review day: what you actually did, new information, statements that did not fit, another possible explanation and the next decision. Do not overwrite the first part; separate the parts by date so the change in understanding stays visible rather than being assembled into an inevitability.
Use phrases such as “I observed” and “not yet known” to leave room for uncertainty. An action should be completable this week and checkable, for example “try a free lesson on Saturday and record lesson and exercise time.” If you never act, record why. Not acting neither confirms nor disproves the advice. On the review day, read the original question first, then the new information, and only then write the decision; that order makes it harder to read the outcome backward into what you supposedly knew.
Saving and exporting: choose a place before a backup
Where a record lives changes what happens next. Guest entries stay in the current browser. Signing in by email saves to your account so you can continue elsewhere; older records are uploaded only when selected. Clearing site data removes browser records not saved to your account, while account history has separate deletion controls. Export important entries as a backup, but remember that an export may contain birth details and private questions; read it before forwarding it or placing it in a shared folder.
When bringing context to an Agent, select only what this question needs. A web result offers “export to Agent,” where you can preview the chart, question and optional background and decide whether to include original birth details. The export does not automatically upload to any Agent. An external Agent's data handling is decided by its host; only the built-in AI feature sends the selected content further to AI. The journal has no note version history, only an editable note field with a length limit, so export before an important edit if you want a record of the earlier version.
Wenbu · reviewing a reading ↗Wenbu · agent integrations ↗Wenbu · current free access ↗
How to read the old record on review day
Start with the original question and first interpretation, then read the action record and new information, and write the revised decision last. If the trial lesson took longer than expected and the exercises took two hours when you have two hours a week, record that conflict and consider “the course pace may clash with my schedule” as an alternative explanation instead of announcing that the cards came true. If a statement never found supporting evidence, write “this was not helpful”; keeping it is more honest than deleting it.
A review is also a good time to check privacy: does this entry still contain unnecessary birth details, other people's information or identifying specifics? Could it keep only the abstract question and observations? Is the export stored somewhere appropriate? These checks do not add predictive value, but they keep the journal usable. The journal does not set reminders for you; put the review date in your own calendar, because the tool will not follow up on a future date.
| Option | When it fits | Deletion and sync notes |
|---|---|---|
| Guest browser journal | Temporary notes or undecided long-term use | Stays in the current browser; clearing site data removes records not saved to an account |
| Signed-in account history | Continuing elsewhere and keeping records together | Older records upload only when selected; account history has separate deletion controls |
| JSON export backup | Important entries, a pre-edit copy or a device change | May contain birth details and private questions; read it before sharing |
| Export to Agent | Showing an external assistant only the relevant context | Preview and choose whether to include original birth details; no automatic upload, and the host handles its own data |
Wenbu · reviewing a reading ↗Wenbu · agent integrations ↗Wenbu · current free access ↗
Hypothetical example: a journal for an eight-week course
This is a fictional practice example for demonstrating the template. It is not a real user observation or a test of predictive accuracy. Suppose someone is considering an eight-week course, knows the fee and does not know the weekly workload.
- September 30, first part: original question, “Should I join an eight-week course? I know the fee, not the workload.” First interpretation, “I want to try a lesson before deciding; I have not enrolled.” Action and date, “Try the free lesson on Saturday and record lesson and exercise time.” Review date, October 7.
- Before saving, remove the course provider's name and personal schedule details, keeping only the decision structure. Choose a browser record and export a JSON backup to a private location you control.
- October 7, second part: “I completed the trial and remain interested. Lesson: 45 minutes. Exercise: 90 minutes. I have two hours a week.” These figures are fictional; a real record should use your own observations.
- What did not fit: “There is no new evidence for the claim about low confidence.” Alternative explanation: “The course pace may conflict with my schedule.” Next decision: “Delay enrollment and try two weeks of shorter exercises.”
- If giving this context to an Agent, select only the question, first interpretation and trial observations, without birth details. Read the export again to confirm it contains no other person's information.
The review asks whether action and new information changed the decision, not whether the cards were “right.” If you never tried the lesson, record why; not acting is not a test of the advice.
Wenbu · reviewing a reading ↗Wenbu · agent integrations ↗Wenbu · current free access ↗
Questions you may have
Does the journal keep versions of a note?
No. It has an editable note field with a length limit and no note version history. If you need the earlier wording, export a backup before an important edit and append new content in a dated paragraph.
Does deleting a local journal entry erase model-provider data?
No. Removing a guest record deletes its browser copy, and account history has separate deletion controls. AI processing, submitted feedback and analytics each follow their own policies, so local deletion is not a full erasure.
Does exporting to an external Agent upload the result automatically?
No. The web export lets you preview the chart, question and optional background and decide whether to include original birth details. An external Agent's data handling is decided by its host; only the built-in AI feature sends selected content further to AI.
Will asking the same question again produce the same tarot card?
Not on demand. A tarot draw is a random procedure that draws without replacement from a 78-card deck, and the returned cards should be kept as recorded. Repeating a question does not guarantee the same card or spread. Deterministic charts such as BaZi can be recalculated from the same inputs; a random draw cannot be reproduced that way.
Wenbu · versioned public capabilities ↗Wenbu · agent integrations ↗
Terms in this guide
- Review record
- A dated comparison of the initial view with newly observed facts, used to separate what was known then from what was learned later.
- Hindsight bias
- The tendency to see an outcome as more foreseeable after learning it; keeping the original record helps reveal that rewriting.
- Alternative explanation
- Another possible reason for an outcome besides the interpretation, including your own actions, schedule or outside conditions.
- Selected context
- The chart, question and background a user previews and chooses to give an Agent; the export does not upload automatically, and the external host handles its own data.
Wenbu · agent integrations ↗Wenbu · versioned public capabilities ↗
See the claim and source notes
A review record should preserve the original question and first interpretation, then add dated observations rather than rewriting the past.
Do not rewrite the first interpretation to make it match the outcome.
Guest journal entries stay in the current browser, and clearing site data removes browser records not saved to an account.
Clearing site data removes browser records that have not been saved to your account.
Local journal notes and JSON export are currently available without payment or registration.
本地手记、JSON 导出
Sources and review scope
This AI-assisted Wenbu article passed automated structure checks and a separate editorial model pass. It is not independent expert review or a paid-account product test.
- Wenbu · reviewing a reading ↗Existing editorial practice · 2026-10-28 · Fetched text
- Wenbu · current free access ↗First-party access and quota documentation · 2026-10-28 · Fetched text
- Wenbu · agent integrations ↗Current first-party product documentation · 2026-10-28 · Fetched text
- Wenbu · versioned public capabilities ↗First-party current tool, convention and privacy contract; not external endorsement or predictive validation · 2026-10-28 · Fetched text
Add a review date to your most recent journal entry
Open a recent entry, append one executable action and a review date, and export a backup before editing; the current journal and export features directly support this.
Add a review date to your most recent journal entry ↗