AnkiWeb addon 226947964
Builds guided native Anki filtered decks from chosen topics and decks, plus recovery, time-based sessions, resume, context, private notes, debriefs, and workload hints.
Open on AnkiWeb GitHub Find alternatives
active
| Min Anki | Max Anki | Updated |
|---|---|---|
| 2.1.1 | + | 2026-08-26 |
Loading…
Displays a post-review summary of cards reviewed, mistakes, and average time, then builds a redo deck of incorrect cards and offers mistake history, leech coaching, and shareable study summaries.
Turns eligible Anki review sessions into a Digimon-inspired RPG where you hatch, raise, digivolve partners, collect items, complete missions, and fight tactical battles and weekly bosses.
Creates a filtered 10-card micro-batch deck, prioritizing due then new cards, and rebuilds it via a customizable shortcut to immediately start review, keeping lapsed cards in the batch.
Provides a keyboard-first search palette for large medical collections with typo/alias matching, native Anki syntax, local semantic search, deck filtering, inline previews, editing, and safe bulk card actions.
Simulates exam conditions through filtered-deck custom study sessions and reports performance metrics including total score, accuracy, time taken and grade.
Builds interactive chess flashcards that each hold a complete line, integrate Anki's grading, support PGN/FEN creation, desktop and iOS, plus syncing shared decks and tactical drilling.
Vellum is an Anki Desktop add-on I made because building a useful filtered deck from a large medical collection had become a chore.
The main feature is Study Builder. It lets you choose topics and then choose which decks those cards should come from. You can also do the reverse: choose a deck first, then browse only the tags that are actually used inside that deck.
Vellum also adds tools for recovering a weak topic, planning a study session by time, resuming interrupted work, leaving private notes for yourself, and seeing what happened after a session.
It does not contain medical cards. It does not write answers for you. It does not replace Anki or FSRS.
Screenshots and a visual walkthrough: https://fateh-alam.github.io/vellum/
Here is the sort of request Study Builder is meant to handle:
Give me Renal or Acid-Base cards from AnKing or my Medicine deck.
You could make that search manually in Anki's Browser. Vellum gives you a guided way to make the same choice without remembering search syntax or digging through the entire tag tree.
Open it from Builder in Anki's top bar, from the Vellum menu, from the Home screen, or from a deck's Overview screen.
Use this route when you already know the subject you want to study.
renal, cardiology, Pathoma, or
high yield.So choosing Renal and Acid-Base from AnKing and Medicine means:
(Renal OR Acid-Base) AND (AnKing OR Medicine)
You do not type that query. It is shown here only to explain exactly what Vellum does with your selections.
Use this route when you know which deck you want to study from, but you do not want to search the collection's full tag list.
Resources > First Aid > Renal.Checking a parent deck includes its subdecks. Opening the arrow beside it only shows the children; it does not select anything. The tag tree works the same way. If you choose a specific child, Vellum removes a redundant selected parent so the scope remains easy to understand.
If you want the whole selected deck scope with no tag filter, use Study entire selected deck scope. Vellum still applies your card-state choice and limit.
Typing in the topic search does not run a fresh collection-wide Anki query for every letter. Vellum searches a local tag/topic index and performs collection work only after the selection changes. Old background results are ignored if you make another choice before they finish.
The final preview separates matching cards into four useful groups:
The card preview is for checking the result. It is not another version of Anki's Browser and it does not offer bulk card editing.
You can choose:
If one side of a due/new percentage is too small, Vellum fills the remaining places from the other side instead of returning a half-empty deck. Sibling avoidance is applied before the final card list is accepted.
Vellum checks the search, counts, ordering, limit, sibling rule, due/new split, and suspension state again immediately before it builds the deck. It does not blindly reuse an old preview.
You can save a setup as a named recipe. The recipe stores the topics, decks, and options you selected. It does not keep cards permanently trapped inside a filtered deck.
Every time you run a recipe, Vellum searches the collection again. Newly added, deleted, moved, retagged, suspended, or unsuspended cards can therefore change the result. If a saved deck or tag no longer exists, Vellum shows it as missing instead of deleting the rule and accidentally broadening the search.
AnKing gets particular attention because it relies heavily on hierarchical tags. Vellum reads the tag structure installed in your own collection. It does not assume that every user has one frozen copy of the V12 hierarchy.
When those paths exist, Vellum can recognize Step containers, systems, subjects, First Aid, B&B, Pathoma, Sketchy, Bootcamp, Pixorize, yield tags, Step overlap, and resource chapter/video paths. Equivalent Step copies can appear as one readable choice while Vellum keeps all of the exact backing tags needed for the real Anki search.
Maintenance, update, and duplicate tags are hidden from the normal topic list so they do not crowd the useful results. They remain available in All tags.
Study Builder is not limited to AnKing. It also works with normal decks and subdecks, flat tags, hierarchical tags, decks without tags, Basic cards, Cloze cards, Image Occlusion cards, and custom note types.
Anki already has a powerful search language, a Browser, Custom Study, and native filtered decks. Study Builder uses those abilities; it does not replace them. What it adds is a guided way to choose topic tags and source decks, a deck-scoped tag tree, an availability check, saved rules, and safer temporary-deck cleanup. If you already enjoy writing Anki searches by hand, you can keep doing that. Vellum is for the cases where finding and combining the right paths is the slow part.
Context is an optional panel beside the card you are reviewing. It reads the card's hierarchical tags and shows where the card sits, for example:
First Aid > Renal > Glomerular disease
This is useful when a card makes little sense in isolation and you need to see the surrounding subject. Depending on your settings, the panel can also show local counts, maturity, retention/failure information, source labels, a Mind Note, and a button to start Recovery for that topic.
Context does not use the question or answer text to guess a topic. Topic parsing comes from decks, tags, and your configured parsing rules.
Anki already shows a card's deck and tags in the Browser and Card Info. Context does not create new metadata. It keeps the useful part of that existing tag hierarchy visible beside the card while you review, and can connect that topic to Recovery or a private Mind Note without making you leave the Reviewer.
Recovery starts from the current card. You choose which level of its topic path needs work, then choose how broad the repair session should be:
Vellum shows the proposed cards before it builds anything. If you continue, it uses Anki's filtered-deck system and normal scheduling. Recovery is for the moment when one failed card tells you the gap is larger than that one fact.
In Anki, you can open the Browser, find the current card's tags, decide how broad the topic should be, write a search, and make a filtered deck yourself. Recovery starts that process from the card in front of you. Its Quick, Standard, Full, and Custom choices decide how wide the topic repair should be, then show the proposed cards before using Anki's normal filtered-deck machinery.
Study by Time is for days when the real limit is time.
Choose how many minutes you have and which deck groups are allowed. Vellum uses answer-time history stored on your computer to estimate how many cards fit. If there is not enough personal timing history yet, it uses conservative fallback times instead of pretending it knows your pace.
The estimate is a planning aid, not a promise that every card will take the same number of seconds. The resulting session is still an ordinary scheduled Anki filtered deck.
Anki can show due counts, limit cards, and report time spent. Its timebox feature can also notify you after a chosen period while you are studying. Study by Time works before the session: you give it minutes and allowed deck groups, and it estimates a card count from your local answer times. It then builds that limited queue with Anki rather than creating a separate study system.
The HUD appears while you study a Vellum session. It can show completed and remaining cards, active and elapsed time, recent pace, the session target, and an estimated time remaining.
It is only a display. It does not decide which answer button you should press and does not alter intervals.
Anki already shows basic remaining counts and records answer time. The Vellum HUD is specific to the session Vellum built. It adds that session's target, active-time estimate, recent pace, and ETA in one small display. These numbers are informational and do not change Anki's queue.
Vellum records checkpoints while you study. If Anki closes, you stop early, or a generated session is interrupted, Resume tells you what it found before you continue.
If the original temporary deck still matches, Vellum can resume that exact session. If the queue has changed because more cards became due, it says so and offers the current context instead of claiming it can restore an identical old queue. You can also abandon an old session deliberately.
Anki normally leaves an unfinished deck available, so you can simply study it again. Resume adds an explanation around Vellum-generated sessions: what you had completed, what was expected to remain, whether the original temporary deck is still the same, and whether newly due cards have changed the situation. It does not duplicate Anki's scheduling state.
Return Ramp can appear after a configurable number of days away. It counts the cards currently available across your active deck groups and separates review, learning, and new cards. It then offers smaller time-based starting sessions.
It does not change the backlog merely because you opened the window. Nothing is built until you choose a session. If an unfinished Vellum deck is already waiting, Return Ramp points you back to Resume or Finish first.
Anki already shows the backlog in its deck counts and lets you use Custom Study. Return Ramp does not hide or reschedule that backlog. It notices that you have been away, puts the current review/learning/new workload in one window, and offers smaller Study by Time sessions as a starting point.
Backtrack lets you look at the previous card and the previous topic context in a separate read-only window. It is useful when you move on and immediately realize you want to compare the last card with the current one.
Viewing the old card does not answer it again, undo anything, or change its scheduling. Backtrack keeps only what it needs for the current Anki run rather than building a permanent archive of card contents.
Anki's Undo command reverses the previous action, and the Browser can preview a card if you go and find it. Backtrack does neither. It opens the previous card and previous topic for reference while leaving the answer and scheduling history alone.
A Mind Note is a short private note attached to a topic, not to a public deck or shared card. You might write:
Compare proximal and distal RTA next time renal physiology comes up.
The note can reappear when the same topic returns. There is also a Mind Notes history where you can review, replace, or clear your notes. This is the one place where Vellum stores prose because it is text you deliberately wrote for yourself. It is stored locally and is not added to the Anki card.
You can already add a field to an Anki note, mark a card, or edit its text. Those changes belong to the collection and may sync or affect a shared note. A Mind Note is separate Vellum data tied to the parsed topic. It is meant for a private reminder to yourself and does not edit the card template or note fields.
After a session, Debrief can show:
A handoff is simply something you want to remember when you return, such as “review nephritic vs nephrotic before adding new renal cards.” The newest handoff can stay on Vellum's Home panel until you mark it done, replace it, or clear it.
Debrief describes what happened in the session. It does not diagnose mastery, predict an exam score, or grade your knowledge.
Anki Stats is the proper place for collection-wide review history, retention, forecasting, and detailed scheduler data. Debrief is much narrower. It explains one study session in ordinary terms and can preserve a one-line reminder for the next time you open Anki. It does not replace Anki Stats or reinterpret FSRS.
The optional Home panel adds Vellum controls to Anki's deck screen. Depending on what you enable, it can show:
Each block can be turned on or off in Settings.
Anki's deck screen already lists decks and their new, learning, and due counts. Vellum leaves that list in place. Its Home panel adds shortcuts and continuity information that Anki does not associate with a Vellum session, such as a saved handoff or an unfinished generated deck.
Topic Health is optional and off by default. It uses recent reviews, failures, and due-card load to sort topics into simple study-attention states. It is meant to answer “where might I need to look next?”
It is a rule-based study hint. It is not a mastery score, medical-knowledge assessment, or prediction of future performance. You can reset its recent-topic data from Settings.
Anki Stats reports real review and scheduling data. Topic Health takes a small set of that local evidence and applies a simple rule to suggest topics that may deserve attention. It is deliberately less ambitious than a statistical model, and you should use Anki's own statistics when you need the underlying numbers.
Vellum keeps local records needed to explain and safely clean up the temporary decks it created. A generated-deck record can include the planned card IDs, original deck IDs, queue positions, answer ratings, answer times, and whether the temporary deck was removed.
It does not copy the front or back of those cards into the history. Old sessions that predate complete tracking are labelled as reconstructed rather than being presented as complete.
Card Info explains the review history of one Anki card. Vellum's generated-deck history explains one temporary session: which card IDs were placed in it, where they came from, what was answered, and whether Vellum removed its deck. It exists mainly for cleanup, recovery, and auditability, not as a replacement for the revlog or Browser.
The first-run guide explains the privacy and scheduling boundaries, lets you choose which modules to enable, and helps Vellum find useful topic roots. You can open it again from Vellum > Getting Started.
Vellum > Commands & Help is an offline guide to the features and available commands. Shortcuts are checked against Anki and other add-ons. If a key is already taken, Vellum disables its conflicting shortcut rather than silently overriding the existing one.
Settings include module switches, topic-root preferences, ignored tag segments, display renames, a parser preview for the current card, timing defaults, absence thresholds, Home-panel choices, history limits, diagnostic export, and separate controls for clearing Vellum data.
Vellum refuses to erase the session/deck records it still needs to clean up an active temporary deck safely.
Anki's add-on manager installs, updates, enables, disables, and configures add-ons. Vellum's own Settings only control Vellum features and data. Its help page explains Vellum commands, and its shortcut check avoids taking a key that Anki or another add-on already uses. You still use Anki's add-on manager to install, remove, or update Vellum.
For Study Builder, Recovery, and Study by Time, Vellum creates a uniquely named native Anki filtered deck. It uses rescheduling mode, so answers made inside that deck update review history and scheduling in the normal Anki/FSRS way.
Vellum records the exact deck identity and the cards Anki placed inside it. When the session finishes, remaining cards return to their home decks and Vellum removes only the temporary deck it owns. An interrupted non-empty deck stays available for Resume.
Only one unfinished Vellum-generated session can exist at a time. If one is already present, Vellum asks you to Resume, Finish, or Abandon it before making another.
Suspended cards stay suspended by default.
Study Builder has an optional Unsuspend and include switch. If you turn it on, Vellum shows the exact number of selected suspended cards and asks for confirmation. It unsuspends only those selected card IDs. They remain active after the session.
If deck creation fails before studying begins, Vellum removes only the newly created deck and re-suspends exactly the cards activated by that failed attempt.
No add-on can honestly promise that it will never have a bug. Keep Anki's normal automatic backups enabled and make occasional manual exports of collections you cannot afford to lose.
Vellum has no account, advertising, telemetry, analytics service, cloud server, API key, or remote crash reporter. It does not make outside network requests.
It does not store card questions, card answers, note fields, media, Study Builder snippets, rendered card content, suggestion counts, or zero-result test queries.
Depending on the modules you enable, its separate database can store card/note/ deck identifiers, timings, answer ratings, parsed topic information, generated- deck manifests, saved recipe selectors, and text you deliberately enter as a Mind Note or handoff.
The diagnostic export removes card and note IDs, private paths, raw tag names, Mind Note text, and handoff text. Check a diagnostic yourself before posting it.
Full data guide: https://github.com/Fateh-Alam/vellum/blob/main/docs/PRIVACY.md
AnkiWeb installations receive updates through Anki's normal add-on update
check. If you install a .ankiaddon file manually from GitHub, that copy is not
connected to the AnkiWeb listing and must be updated manually.
Vellum 1.1.1 targets Anki Desktop 25.09.4 or newer.
The 1.1.1 release was tested on macOS with Anki 25.9.4 and 26.5. The source has Windows support, but the full Windows 11 test list has not yet been completed. Linux has not been tested live for this release. I do not claim compatibility with every third-party add-on.
The release checks included 162 unit tests, Python and JavaScript checks, package integrity checks, a 50,000-tag search test, temporary Anki collections for deck creation/scheduling/rollback tests, light and dark Qt interface checks, and real Anki 26.5 startup checks on macOS.
Testing details: https://github.com/Fateh-Alam/vellum/blob/main/docs/DEVELOPMENT.md
Source and release downloads: https://github.com/Fateh-Alam/vellum
Bug reports, collection-safety concerns, and feature requests: https://github.com/Fateh-Alam/vellum/issues/new/choose
General feedback and questions: https://github.com/Fateh-Alam/vellum/discussions
Private security or privacy reports: https://github.com/Fateh-Alam/vellum/security/advisories/new
Vellum is not affiliated with Anki, AnkiHub, or AnKing.
Anki remembers your cards. Vellum remembers your studying.
See the visual walkthrough · Install from AnkiWeb · Download the latest release · Give feedback
AnkiWeb code: 226947964
I built Vellum because I was spending too much time in Anki's Browser trying to combine medical topics, resource tags, parent decks, and subdecks. The cards were already in the collection; getting the exact set I wanted was the annoying part.
Vellum is an Anki Desktop add-on that handles that setup. Its main feature, Study Builder, turns a request such as:
I want Renal or Cardiology cards, but only from AnKing or my Medicine deck.
into a scheduled, rescheduling-enabled native Anki filtered deck—without making you construct Browser queries.
Vellum does not provide medical content, generate cards, replace FSRS, choose answer ratings, edit your tags, or upload your collection. Anki remains the owner of your cards and scheduling.
Your existing cards
↓
Choose a topic/deck scope you understand
↓
Inspect ready, suspended, buried, and unavailable cards
↓
Vellum creates a temporary native filtered deck
↓
You answer normally with Anki and FSRS
↓
Cards return to their home decks when the session is safely finished
In other words: Vellum picks the cards; Anki still handles the studying.
Use this when you know the subject you want to study.
Search only topics and tag paths—for example renal.
Select one or several topics. Multiple topics are alternatives: Renal OR Acid–Base.
Vellum shows only decks and subdecks that actually contain matching cards.
Choose one or several source decks. Multiple source decks are alternatives: AnKing OR Medicine.
Vellum combines the two sets:
(Renal OR Acid–Base) AND (AnKing OR Medicine)
Inspect exact availability and a few card previews.
Choose a limit and select Build & Study.
Use this when you know which deck should supply the cards.
| Bucket | Meaning |
|---|---|
| Ready | The card can enter this filtered deck now. |
| Suspended | It remains excluded unless you explicitly enable and confirm unsuspension. |
| Buried | Vellum reports it, but it cannot enter the filtered deck. |
| In another filtered deck | Vellum reports it and refuses to steal it from that deck. |
Preview snippets are inspect-only and ephemeral. They are not stored in Vellum's database, recipes, logs, diagnostics, or Debriefs.
No software can honestly promise it will never contain a bug. Vellum is designed with a narrow mutation boundary, exact ownership checks, compensating rollback, and disposable-collection qualification. Keep Anki's normal automatic backups and make occasional manual exports of important collections.
user_files/ directory.Immediately before building, Vellum re-evaluates the definition, counts, ordering, limit, sibling policy, due/new share, and activation eligibility.
If filtered-deck creation fails before study begins, Vellum removes only the newly created identity-checked deck and re-suspends exactly the cards activated for that attempt. It will not delete a deck when the saved ID/name ownership no longer matches.
Natural completion or explicit Finish returns remaining cards home and removes the exact Vellum-owned temporary deck. An interrupted, non-empty deck remains resumable. Only one unfinished Vellum-generated study session may exist at once; Vellum offers Resume, Finish, or Abandon instead of quietly creating overlapping sessions.
AnKing is primarily organized with hierarchical tags because a card can relate to several systems, resources, and exams while living in only one deck. Vellum detects the locally installed hierarchy instead of assuming one frozen V12 tag snapshot.
When present, it exposes human-facing structures such as:
Parent selections include descendants. Deterministically equivalent Step copies can appear as one visible row while their exact raw backing tags are preserved for compilation and recipes. Maintenance/update/duplicate tags stay out of the ordinary view but remain discoverable under All tags.
Vellum also works with ordinary parent/subdecks, generic hierarchical tags, flat tags, decks without tags, Basic, Cloze, Image Occlusion, and custom note/card types.
Every module can be disabled independently. Topic Health is opt-in and explicitly labeled as a workflow heuristic—not a mastery score.
Vellum has no account, telemetry, analytics SDK, cloud backend, API keys, or external data requests.
It never stores card question/answer text, note fields, media, Study Builder snippets, rendered content, suggestion counts, or zero-result diagnostic probes. It may store identifiers, timings, answer ratings, topic metadata, exact generated-deck manifests, user-selected recipe selectors, and text deliberately written in a Mind Note or handoff.
Diagnostics redact card/note IDs, profile paths, tag names, Mind Note text, and handoff text. Read the complete privacy and data guide.
226947964.AnkiWeb installations are offered updates through Anki's add-on updater.
vellum-1.1.1.ankiaddon from the
latest release.A manually installed GitHub file is not attached to the AnkiWeb listing and must be updated manually.
Vellum 1.1.1 targets Anki 25.09.4 or newer and was locally qualified on macOS with Anki 25.9.4 and 26.5. Native Windows-compatible source is present, but the real Windows 11 matrix remains pending. Linux is not live-qualified for 1.1.1.
Open Vellum from Anki's Builder top-bar item, Vellum → Build Study Deck…,
Home, Overview, or the default Ctrl+Shift+B shortcut. The shortcut is
conflict-checked and displayed using your operating system's native notation.
The one-time Getting Started guide explains privacy and scheduler boundaries, lets you choose modules atomically, detects topic roots, and shows actual workflow entry points. Reopen it with Vellum → Getting Started…. The complete offline guide is under Vellum → Commands & Help….
The v1.1.1 qualification run included:
These checks do not prove compatibility with every third-party add-on or another operating system and are not presented as such. See Development and release checks.
Vellum's core, builder, topic, Recovery, session, and timing logic remains importable without Anki. Qt and WebView dependencies stay at the integration edge.
python3 -m unittest discover -s tests -t .. -v
env PYTHONPYCACHEPREFIX=/tmp/vellum-pycache python3 -m compileall -q .
node --check ui/assets/vellum.js
python3 scripts/performance_smoke.py
python3 scripts/build_package.py
python3 scripts/release_check.py
The Anki-core and Qt scripts must be run from outside the repository using Anki's bundled Python and disposable collections.
Read Troubleshooting first. If the issue remains, export a redacted diagnostic from Vellum → Settings → Data, check it, and choose the right place to respond:
Do not attach your collection, card content, Vellum database, Mind Notes, handoffs, or raw profile paths. Report security/privacy vulnerabilities through GitHub's private vulnerability-reporting form.
Vellum is not affiliated with Anki, AnkiHub, or AnKing.