Open data
How the library works
Adding, reviewing, correcting and translating entries, in plain language.
The QuiJutsu Anime Asset Library is the catalogue of characters, places, objects, abilities and everything else QuiJutsu's trivia questions are built from. Players build it: anyone signed in can add what is missing, check what others added, correct a published entry or translate one.
Nothing goes live on one person's word. Every entry and every change is checked by other players, waits a day in the open, and can still be reported after it publishes.
Four ways to help
Add something new
A character, place, object or anything else the library does not have yet. You give it a name, any other names it goes by, a few details and where it can be seen.
Review other people's entries
Open an entry from the review queue and check it section by section, laid out the way it was filled in. Nothing publishes without reviewers.
Correct a published entry
Use Suggest a change on the entry. The form opens with the published values, you fix what is wrong and say why, and only the fields you changed go to review.
Translate an entry
Use Translate into on a published entry to write its name, other names and details in another language. Members who read that language check it.
From submission to the library
1. You submit it
Pick the series and the kind, then give it a name, any other names, and details: short statements such as “Leader of the Hero Party”, one per row. Rate how big a deal it is, from Rookie to Godly, and say how anyone can check it: an episode and timestamp on an official stream, a volume and chapter if it was never animated, or that it is well known enough to need no source.
The library checks first whether it already exists under that name or another name it is known by, so it never creates a duplicate.
2. Reviewers check each field
Other players open it on its own page, laid out like the form you filled in. Nobody can review their own entry. For the kind, the name and the evidence level, a reviewer agrees with what was submitted, backs another reviewer's suggestion, or suggests a correction.
Other names are kept or dropped one by one, and each detail gets a tick from a reviewer who can stand behind it. The reviewer also gives their own Spirit rating. Every section has to be marked before Approve is available.
Reviews have to be independent. If someone has approved a lot of one contributor's work recently, their approvals stop counting on that contributor's entries, so a small group cannot wave each other's entries through.
3. 4 approvals settle it
An entry needs 4 reviewers to approve it, whatever its evidence level, and 2 who reject it are enough to refuse it. On an entry that names a source, each approver opens it and says which one they checked. On a well-known entry, approving means you recognise it without one.
Fields settle the same way. A suggested correction replaces the submitted value once 4 reviewers back it, and a newly added other name needs 4 too. A detail publishes once 4 reviewers tick it, the submitter's own details included. A detail that does not get there is left out and does not hold the entry back.
4. A 24-hour window
Settling does not publish it. The entry waits 24 hours on Publishing Soon, where any signed-in member can see it and who approved it. Anyone with standing can challenge it in that time.
5. Published
When the window closes without a challenge, the entry joins the public library and can be used in trivia questions. It can still be corrected, translated or reported afterwards.
Evidence levels
You pick one when you submit, and reviewers can move it. The bar is the same for all three. What changes is what a reviewer checks and who may review.
| Evidence level | What a reviewer checks | Approvals | Reviewer level |
|---|---|---|---|
| Well known | That they recognise it without a source | 4 | 1 |
| It was animated | The episode and timestamp | 4 | 1 |
| Manga only | The volume, chapter and page | 4 | 2 |
An approval that names a source counts towards a well-known entry too, so one reviewer who finds the episode and three who recognise it settle it. Manga-only entries need reviewers at trust level 2, because far fewer people can check print than a stream. While the library is being founded, reviewers at any level count on every route.
A few examples
A famous main character everyone recognises
Nobody needs to find a timestamp, but nobody gets to skip the check either. Four reviewers agree it is well known, the window passes without a challenge, and it publishes. It skips the source, not the agreement.
Someone else finds the source for your entry
You add a character but cannot place the episode. A reviewer tracks it down and names it while approving, and earns double what a plain approval pays. Finding the source is the hard part, so it is paid that way.
A minor weapon from episode 14
You submit it with the episode and the timestamp where it appears. Each reviewer opens that moment, sees it is really there and approves. Naming what they checked is what makes the approval a check rather than a guess.
A character who only appears in the manga
Never animated, so there is no timestamp to give. The evidence is a volume, chapter and page instead, and reviewers need trust level 2, because far fewer people can check a printed page than a streamed episode.
A detail nobody can confirm
You add three details to a character. Reviewers tick the two they can stand behind but not the third. The entry publishes with two details, and the third is simply left out. It does not hold the rest back.
Someone spots a mistake during the window
An entry settles, but during the window a reviewer notices the name belongs to a different series and challenges it. Its approvals are cleared and it goes back to the queue, where reviewers who have not seen it decide. If they settle it again it gets a fresh window. If it is challenged a second time, a curator decides.
A published name is misspelt
You open the entry, choose Suggest a change, fix the spelling and say why. Only the name goes to review, labelled Change in the queue. Until 4 reviewers approve it and the window passes, the library keeps showing the old spelling.
Translating an entry with one detail you are unsure of
You translate the name and two details and leave the third blank. Readers of that language back or improve each part, and each one publishes on its own at 3 backers. The blank detail keeps showing in English until someone translates it.
A brand new account checking a famous character
New accounts start at trust level 0, and their approvals do not count yet. They can still review as an apprentice: the mark is recorded and builds their own track record, it just does not settle the entry. It is a safe way to start before you have earned full standing.
Challenging something before it publishes
A challenge needs a written reason, and it stops the clock. The entry leaves its window, its approvals stop counting and it returns to the queue. Each person reviews an entry once, so everyone who reviewed it the first time is locked out, and it can only settle again on reviewers who have not looked at it yet.
That happens once. If an entry survives a challenge and someone challenges it again, a curator decides instead. There is no third round, so a challenge and a fresh approval can never loop against each other.
You can have 3 challenges open at a time. That is enough to be useful and not enough to stall the queue on your own.
After it publishes
Anyone who can review can correct a published entry with Suggest a change. The form opens with the published values and asks why you are changing them. Only the fields you change are reviewed: the name, other names, details or evidence level. A change needs 4 approvals and goes through the same window and challenge rules as a new entry, and the published values stay until it settles.
Report this entry is for an entry that should not exist at all. If a curator agrees, it is hidden from the public library rather than deleted, so its history stays, and the approvals that put it there count as overturned. You can have 5 reports open at a time and one per entry.
Translate into lets you write a published entry's name, other names and details in another language. Leave a part blank and it keeps showing in English. Members who list that language under Languages you read in Settings review it: they agree, back another suggestion or suggest their own, and each part publishes at 3 backers.
Changes and translations wait in the same review queue as new entries, labelled Change and Translation.
Rating how big a deal it is
The submitter rates how well known or central the entry is to its series, and every reviewer gives their own rating. Once it publishes, all of those ratings are averaged into one Spirit level:
After that, players can vote on it in public. At 10,000 votes the level is certified and stops moving.
Trust levels, and how to move up
Your trust level decides which entries your approval counts on. It is based on accuracy, not activity: only reviews of entries that reached a final outcome count, and only when your review matched it.
| Level | How to reach it |
|---|---|
| 0 | Everyone starts here. You can submit entries and review as an apprentice: your marks are recorded but do not settle anything. |
| 1 | 5 of your own entries accepted, or 10 reviews that matched the final outcome, at least 5 of them from the review queue. Either path works alone. |
| 2 | 25 or more accepted entries, at least 20 settled reviews, and 90% or more of those reviews matching the final outcome. This is the level manga-only entries need. |
Being wrong has a cost. If something you approved is later reported and a curator agrees it was wrong, your approval counts as overturned and against your accuracy. That keeps a high level meaningful.
Rewards
Coins and XP are paid when an entry is finally settled, and only to the people whose review matched the result. If it publishes, the submitter and every approver are paid. If it is refused, only those who rejected it are. Being right pays, not clicking.
Finding the source for someone else's entry pays double a plain approval. Reviewers who take the second look after a challenge earn a little more, and a challenge that turns out to be right pays too.
Milestones earn badges on four tracks: entries you added, sources you found, checks that held up, and wrong entries you caught. They are worked out live from your record, so a badge can disappear if something you approved is later overturned.
When a challenge turns out to be wrong
Being wrong about a challenge costs you nothing. Someone who stops an entry in good faith and turns out to be mistaken has still made others look again, and charging for that would produce a library nobody checks. What gets recorded is a flag, and three flags within 90 days put the case in front of a curator.
Three flags are an accusation, not a finding. A curator reads the challenges side by side and can simply clear them, which is the expected answer when someone was careful and unlucky. If the pattern is deliberate, the result is a review ban: seven days, then thirty, then indefinite. It removes the review queue and nothing else, and it can be appealed.