A Typst package for creating horizontal columned lists, similar to LaTeX’s tasks package. Perfect for exercises, multiple-choice questions, and any content that benefits from a compact columned layout.
Gallery
Click on an image to see the source code.
| Basic Tasks | Math Exercises | Custom Labels | Styled Layout |
| Vertical Flow | Column Spanning | Bold Labels | Resume Numbering |
| Adaptive Row Heights | Wrap Zone | Auto-Fit Columns |
Features
- Indentation-independent parsing - Output is completely independent of source code indentation
- Horizontal flow layout - Items flow left-to-right across columns (like LaTeX’s tasks)
- Flexible column counts - 2, 3, 4 or any number of columns
- Auto-fit columns - Use the largest column count that does not introduce a new wrap or overflow
- Column spanning - Items can span multiple columns using
(N)syntax - Multiple label formats - Alphabetic (
a),A)), numeric (1),(1)), roman (i),I)), bullets, or custom - Bold labels - Make labels stand out with bold weight
- Resume numbering - Continue numbering across multiple task blocks
- Two flow directions - Horizontal (a b | c d) or vertical (a c | b d)
- Global configuration - Set defaults for all tasks in your document
- Shorthand functions -
tasks2,tasks3,tasks4for quick column setup - Spacing & alignment controls - Fine-grained gutters, label alignment, baselines, and block spacing
- Wrap zone - Reserve a top-right corner for overlay content (QR codes, stamps) and flow task rows around it
Manual
A full manual is available as a PDF, with a Typst source version in docs/manual.typ.
Quick Start
#import "@preview/taskize:0.2.12": tasks
#tasks[
+ First item
+ Second item
+ Third item
+ Fourth item
]
Basic Usage
Simple Two-Column Layout
#import "@preview/taskize:0.2.12": tasks
#tasks[
+ $2 + 3 = ?$
+ $5 - 2 = ?$
+ $4 times 3 = ?$
+ $8 div 2 = ?$
]
Three-Column Layout
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: 3)[
+ Option A
+ Option B
+ Option C
+ Option D
+ Option E
+ Option F
]
Auto-Fit Columns
Use columns: "auto-fit" to choose the largest column count that does not make any item wrap or overflow compared with a one-column layout. Limit the search with max-columns.
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: "auto-fit", max-columns: 4)[
+ Medium length task with detail
+ Short
+ Short
+ Short
]
By default, auto-fit uses auto-fit-mode: "fill": an item too wide for one
column automatically spans the minimum number of columns it needs — no
explicit +(N) required — so wide items share rows with densely packed short
items. Use auto-fit-mode: "uniform" when every item must fit one ordinary
column; in that mode, the column count is the largest value where no item
wraps at one-column width, so a wide item can force the whole block to fewer
columns. If an item already wraps in the one-column layout, auto-fit keeps
one column, because no no-wrap multi-column layout exists.
Auto-fit has to lay an item out to measure it, and an item that draws
itself (a CeTZ canvas, a plot) is computed again each time. Text and inline
math are judged on their natural width alone, and a block or box given
both a fixed width and a fixed height is not laid out at all: its size is
read from the call. Wrap an expensive figure in one when you know its size.
#tasks(columns: "auto-fit", max-columns: 4)[
+ #block(width: 4cm, height: 3cm, figure-a)
+ #block(width: 4cm, height: 3cm, figure-b)
]
Settings passed to tasks itself (or once with
#let tasks = tasks.with(columns: "auto-fit")) are known on the first layout
pass. Settings made with tasks-setup are only known on the second one, so a
document full of figures compiles faster with the former.
Shorthand Functions
#import "@preview/taskize:0.2.12": tasks2, tasks3, tasks4
// Two columns
#tasks2[
+ Item 1
+ Item 2
]
// Three columns
#tasks3[
+ Item 1
+ Item 2
+ Item 3
]
// Four columns
#tasks4[
+ A
+ B
+ C
+ D
]
Label Formats
Built-in Formats
#import "@preview/taskize:0.2.12": tasks
// Lowercase letters with parenthesis (default)
#tasks(label: "a)")[+ One + Two + Three]
// Uppercase letters
#tasks(label: "A)")[+ One + Two + Three]
// Numbers with parenthesis
#tasks(label: "1)")[+ One + Two + Three]
// Numbers in parentheses
#tasks(label: "(1)")[+ One + Two + Three]
// Lowercase roman numerals
#tasks(label: "i)")[+ One + Two + Three]
// Uppercase roman numerals
#tasks(label: "I)")[+ One + Two + Three]
// Bullet points
#tasks(label: "*")[+ One + Two + Three]
// Dashes
#tasks(label: "-")[+ One + Two + Three]
// No labels
#tasks(label: none)[+ One + Two + Three]
Custom Label Function
#import "@preview/taskize:0.2.12": tasks
// Custom emoji labels
#tasks(label: n => "Q" + str(n) + ":")[
+ What is 2+2?
+ What is the capital of France?
+ What color is the sky?
]
Flow Direction
Horizontal Flow (Default)
Items fill rows first: a b | c d | e f
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: 2, flow: "horizontal")[
+ a
+ b
+ c
+ d
+ e
+ f
]
// Layout:
// a) b)
// c) d)
// e) f)
Vertical Flow
Items fill columns first: a c e | b d f
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: 2, flow: "vertical")[
+ a
+ b
+ c
+ d
+ e
+ f
]
// Layout:
// a) d)
// b) e)
// c) f)
Resuming Numbering
#import "@preview/taskize:0.2.12": tasks, tasks-reset
#tasks[
+ First
+ Second
+ Third
]
Some text between task blocks...
// Continue numbering from where we left off
#tasks(resume: true)[
+ Fourth
+ Fifth
+ Sixth
]
// Reset counter to start fresh
#tasks-reset()
#tasks[
+ Back to one
+ Two again
]
Starting from a Specific Number
#import "@preview/taskize:0.2.12": tasks
#tasks(start: 5)[
+ This is item 5
+ This is item 6
+ This is item 7
]
Global Configuration
Set defaults for all tasks in your document:
#import "@preview/taskize:0.2.12": tasks, tasks-setup
// Configure global defaults
#tasks-setup(
columns: 3,
label-format: "1)",
column-gutter: 1.5em,
row-gutter: 0.8em,
)
// All subsequent tasks use these defaults
#tasks[
+ Item 1
+ Item 2
+ Item 3
]
// Override specific settings per-call
#tasks(columns: 2, label: "a)")[
+ Override A
+ Override B
]
Advanced Features
Label Baseline Alignment
Control the vertical alignment of labels relative to content:
#import "@preview/taskize:0.2.12": tasks
// Center alignment (default)
#tasks(label-baseline: "center")[
+ Short item
+ Multi-line item\
with more text
]
// Top alignment
#tasks(label-baseline: "top")[
+ Item aligned at top
+ Another item
]
// Bottom alignment
#tasks(label-baseline: "bottom")[
+ Item aligned at bottom
+ Another item
]
Bold Labels
Make labels stand out by using bold weight:
#import "@preview/taskize:0.2.12": tasks
// Bold labels for emphasis
#tasks(label-weight: "bold")[
+ Important task
+ Critical item
+ Key point
]
// Regular labels (default)
#tasks(label-weight: "regular")[
+ Normal item
+ Standard task
]
Column Spanning
Items can span multiple columns using the +() or +(N) syntax (no space after +). This is similar to LaTeX’s \task* command:
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: 3)[
+ Short
+ Short
+ Short
+(2) This item spans 2 columns
+ Short
+() This item spans all columns (full width)
+ Normal
+ Normal
]
// Practical example: Multiple choice with explanation
#tasks(columns: 2, label: "A)")[
+ Paris
+ London
+ Berlin
+ Madrid
+() The capital of France is Paris, known for the Eiffel Tower.
]
Syntax:
+ Content- Normal item (spans 1 column)+(2) Content- Spans 2 columns (no space after +)+(3) Content- Spans 3 columns+() Content- Spans all columns (full width)
Notes:
- Important: No space between
+and()or(N) - If an item with span doesn’t fit in the current row, it wraps to the next row
- Column spans are clamped to the total number of columns
- Great for adding explanations, headings, or long content within lists
Answer Lines
Room to write in, under each task — the lines fill the column they land in, so they follow the layout instead of a fixed width.
#import "@preview/taskize:0.2.12": tasks
#tasks(columns: 2, answer-lines: 2)[
+ Calculate $2 + 2$
+ Calculate $3 times 5$
]
answer-line-style picks how they are drawn — "dotted" (default),
"dashed", "solid", or any stroke, e.g. (paint: gray, thickness: 0.6pt).
answer-line-gap sets the distance between two lines, and
answer-line-above / answer-line-below the distance to the task above and
to whatever follows (auto leaves Typst’s own block spacing in place).
#tasks(
columns: 2,
answer-lines: 3,
answer-line-style: "solid",
answer-line-gap: 1.6em,
answer-line-above: 0.8em,
answer-line-below: 1.4em,
)[
+ Solve $x^2 = 9$
+ Expand $(x + 1)^2$
]
All of them are settable document-wide with tasks-setup. To draw lines
outside a task list, answer-dots takes the same options:
#answer-dots(3)
#answer-dots(2, style: "solid", gap: 1.4em, above: 0.8em)
Each line fills the cell it lands in rather than a fixed length, so the
lines follow whatever layout the block ends up with: an item spanning two
columns gets a line twice as long, columns: "auto-fit", flow: "vertical",
row-gutter: "auto", a nested tasks call and a wrap-zone all keep their
own widths, and inside Typst’s own #columns() the lines fit the text
column. Note that the lines are drawn under every item, a
+ () instruction spanning the full row included — keep such a heading
outside the tasks call if it should not get one.
Parameters Reference
tasks Function
| Parameter | Type | Default | Description |
|---|---|---|---|
columns |
int/string | 2 | Number of columns, or "auto-fit" to choose the largest no-wrap count |
label |
string/function | "a)" |
Label format (shorthand for label-format). "enum" follows the document’s enum numbering, one level per nesting depth |
label-format |
string/function | "a)" |
Label format |
answer-lines |
int | 0 | Ruled lines drawn under each task, for exam sheets |
answer-line-gap |
length | 1.2em | Spacing between those lines |
answer-line-style |
string/stroke | "dotted" |
"dotted", "dashed", "solid", or a stroke to draw the rules with |
answer-line-above |
length/auto | auto | Space between the task and its first answer line |
answer-line-below |
length/auto | auto | Space after the last answer line |
start |
int | 1 | Starting number for labels |
resume |
bool | false | Continue numbering from previous block |
column-gutter |
length | 1em | Space between columns |
row-gutter |
length/string | 0.6em | Space between rows. "auto" (or "adaptive") sizes rows to the true ink of their content so tall inline math (display fractions, matrices) no longer bleeds into the gutter |
max-columns |
int/auto | auto | Maximum column count tested by columns: "auto-fit"; auto means up to the number of items |
auto-fit-mode |
string | "fill" |
"fill" measures spans at their rendered width; "uniform" requires every item to fit one ordinary column |
auto-fit-tolerance |
length | 0.5pt | Measurement tolerance for auto-fit wrap detection |
label-width |
auto/length | auto | Width reserved for labels |
label-align |
alignment | right | Label alignment |
label-baseline |
string/length | "center" |
Label vertical alignment: "center", "top", "bottom", or length |
label-weight |
string | "regular" |
Label font weight: "regular" or "bold" |
indent-after-label |
length | 0.4em | Space between label and content |
indent |
length | 0pt | Left indentation of entire block |
above |
length | 0.5em | Space before block |
below |
length | 0.5em | Space after block |
flow |
string | "horizontal" |
Flow direction: "horizontal" or "vertical" |
Label Format Options
| Format | Example | Description |
|---|---|---|
"a)" |
a) b) c) | Lowercase letters with closing paren |
"a." |
a. b. c. | Lowercase letters with period |
"(a)" |
(a) (b) © | Lowercase letters in parentheses |
"A)" |
A) B) C) | Uppercase letters with closing paren |
"A." |
A. B. C. | Uppercase letters with period |
"1)" |
1) 2) 3) | Numbers with closing paren |
"1." |
1. 2. 3. | Numbers with period |
"(1)" |
(1) (2) (3) | Numbers in parentheses |
"i)" |
i) ii) iii) | Lowercase roman numerals |
"I)" |
I) II) III) | Uppercase roman numerals |
"*" |
bullet | Bullet points |
"-" |
dash | En-dashes |
none |
(none) | No labels |
| function | custom | Custom function n => content |
Examples
Math Exercise Sheet
#import "@preview/taskize:0.2.12": tasks
= Algebra Practice
Simplify the following expressions:
#tasks(columns: 2)[
+ $x^2 + 2x + 1$
+ $(a + b)^2$
+ $3x - 5 + 2x + 8$
+ $frac(x^2 - 4, x - 2)$
]
Solve for $x$:
#tasks(columns: 2, label: "1)")[
+ $2x + 5 = 13$
+ $x^2 = 16$
+ $3(x - 2) = 15$
+ $frac(x, 4) = 7$
]
Multiple Choice Questions
#import "@preview/taskize:0.2.12": tasks
*Question 1:* What is the capital of France?
#tasks(columns: 2, label: "A)")[
+ London
+ Paris
+ Berlin
+ Madrid
]
*Question 2:* Which planet is closest to the Sun?
#tasks(columns: 2, label: "A)")[
+ Venus
+ Mercury
+ Mars
+ Earth
]
Vocabulary List
#import "@preview/taskize:0.2.12": tasks
= French Vocabulary
#tasks(columns: 3, label: none)[
+ *bonjour* - hello
+ *merci* - thank you
+ *oui* - yes
+ *non* - no
+ *s'il vous plait* - please
+ *au revoir* - goodbye
]
When to Use taskize
taskize is specifically designed for horizontal columned layouts where items flow left-to-right across multiple columns, perfect for exercises, multiple-choice questions, and compact lists. Choose taskize when you need:
- Horizontal flow layout - Items arranged in rows across columns (a b | c d | e f)
- Compact exercise sheets - Mathematical exercises, vocabulary lists, multiple-choice questions
- Flexible column control - Easy 2, 3, 4+ column layouts with shorthand functions
- Resume numbering - Continue numbering across multiple task blocks
- LaTeX tasks compatibility - Familiar syntax for users migrating from LaTeX’s tasks package
Comparison with itemize
itemize is an excellent, feature-rich package for vertical list styling with advanced customization capabilities. Use itemize when you need:
- Vertical lists with advanced styling (colors, custom markers, spacing)
- Nested lists with sophisticated formatting
- Rich marker customization (emojis, custom content, per-item markers)
- Advanced typography control over list appearance
- Complex list hierarchies with fine-grained control
In summary: Use taskize for horizontal multi-column layouts (exercises, quizzes, compact lists), and use itemize for feature-rich vertical lists with advanced styling. They complement each other well for different layout needs.
License
MIT License - see LICENSE file for details.
Changelog
All notable changes to taskize are documented here.
[0.2.12] - 2026-11-17
Fixed
- Items of a row now share one baseline. Grid cells are top-aligned, so an item whose first line is tall (a display fraction, a vector) sat lower than a plain neighbour:
c) 8x - 8 = -7x + 7stood abovea)andb)when those started with fractions. The first line of each item is measured and the shorter ones are lowered, so the first lines of a row line up, for inline items and for items that start inline and continue with a block. - The label takes no room in the first line. It is set in a zero-width box at the start of that line, as was already done for items continuing with a block. An unbreakable formula almost as wide as the column now stays beside its label instead of dropping below it and leaving the label alone on its line. Wrapped lines keep the same indent, so the layout is otherwise unchanged.
[0.2.11] - 2026-11-10
Changed
columns: "auto-fit"lays items out far less. Column counts are tried from the largest down, after a pass on natural widths alone that discards the counts an item is plainly too wide for. Text and inline math are judged on their natural width without any further layout, and nothing is measured when a single column is all there can be. The column counts and spans chosen are the ones 0.2.10 chose (checked on several thousand generated lists, in both modes); a document of short items compiles faster and holds less memory.- A
blockorboxwith a fixedwidthandheightis not measured at all by auto-fit: its size is read from the call. Measuring a figure that draws itself (a CeTZ canvas, a plot) costs as much as drawing it, on every layout pass, so wrapping such a figure in a fixed-size container brings auto-fit down to the cost of a fixed column count.
Added
answer-line-style,answer-line-above,answer-line-belowontasks,tasks-setupandanswer-dots(asstyle,above,below): answer lines can be dotted, dashed, solid or drawn with any stroke, and their distance to the task and to what follows is adjustable.answer-rule, a single answer line filling the width it is given.
[0.2.10] - 2026-09-01
Fixed
columns: "auto-fit"overflowed rigid content — tables, fixed-size figures and images. A table with fixed columns keeps exactly the same height however narrow the column it is measured in: it does not wrap, it overflows. Auto-fit judged items by height alone, so every column count looked like a fit and it chose the densest one, laying tables and fixed-size graphics on top of each other. Content whose height does not change at a narrower width is now also measured by its natural width, and gets the columns it needs — spanning them in"fill"mode, lowering the count for the block in"uniform"mode. Content that does adapt is unaffected: text and table cells still wrap, and an image with no explicit width still scales down to keep its column count. Reported for ratio tables in exercise sheets.
Added
- Gallery example for auto-fit (
gallery/auto-fit.typ) and a manual section on how rigid content is measured.
[0.2.9] - 2026-08-19
Added
label: "enum"(#3) — follow the document’s ownenumnumbering instead of repeating the format on every call. A multi-level pattern is honoured by nesting depth: with#set enum(numbering: "1) a)"), a top-level task list numbers1)2)and a nested onea)b). The default is unchanged.answer-lines(#1) — dotted lines under each task for exam sheets, filling the column they land in:#tasks(columns: 2, answer-lines: 2)[...].answer-line-gapsets their spacing, andanswer-dots(count: 3)draws them anywhere on its own. Settable document-wide withtasks-setup.- Items as an array —
#tasks(([One], [Two], [Three]))spells a list out where the one-line+syntax cannot; an entry may carry its span as([wide], 2).
Fixed
- A bare
+inside an item was taken for a new item (#2) —+ What is 2 + 2?was split in two and the rest of the line repeated after it. Typst only starts an item at the head of a line, so a+anywhere else is plain text and is now treated as such. A list written on one line has no item syntax to rely on, so pass the items as an array instead:#tasks(([One], [Two], [Three])), which also takes([wide item], 2)to span columns. - The label of an item that opens inline and continues with a block sat at the top of the cell instead of on the item’s first baseline. An item written as a line of text followed by a nested
taskscall, a figure or a table is not inline as a whole, so it fell back to the top-aligned rendering: with a display fraction on that first line, the number ended up level with the numerator rather than with the text it labels. This is the ordinary shape of a multiple-choice question, and it shows on every item when fractions are set in display style. The item is now indented by the label column and the label pulled back into the margin that frees, so it rides on the first line and shares its baseline. The content itself is passed through untouched, so a nestedtaskskeeps its ownabovespacing exactly. Items that begin with a block keep the previous two-cell rendering, there being no text line to align on. Reported, with this approach, by the MathALÉA team.
[0.2.8] - 2026-07-17
Added
Fixed - Reserve a rectangular zone at the top right of a #tasks call with the wrap-zone parameter (or the shared taskize-wrap-zone state) so task rows flow around overlay content placed there independently — a QR code, a stamp, a logo. Rows overlapping the zone render narrowed beside it; later rows return to full width once the zone’s height is cleared. The shared state is the cross-package contract wrapper packages (e.g. exercise-bank) use to reserve space for content they draw themselves.
- Adaptive row heights -
row-gutter: "auto"(or"adaptive") sizes rows to the true ink of their content, so tall inline math (display fractions, matrices) no longer bleeds into the row-gutter and the configured numeric gutter becomes the real visual gap. Rows with ordinary content keep exactly the same spacing as before. Can be set document-wide withtasks-setup(row-gutter: "adaptive").
[0.2.7] - 2026-07-13
Added
- Auto-fit columns -
columns: "auto-fit"selects the largest column count, bounded bymax-columns(default: up to the number of items), that does not introduce a new item wrap or fixed-size content overflow. Useful for exercises mixing short choices, longer text, math, spans, and small figures. - Auto-spanning in fill mode - with
auto-fit-mode: "fill"(default), a regular+item too wide for one column automatically spans the minimum number of columns it needs — no explicit+(N)required — so wide items share rows with densely packed short items.auto-fit-mode: "uniform"instead requires every item to fit one ordinary column and falls back to fewer columns when the widest item no longer fits.
[0.2.6] - 2026-07-02
Added
- True baseline alignment — with
label-baseline: "center"(default), inline content (text, inline math, symbols) now shares the same paragraph line as the label via hanging-indent layout, so fractions and tall inline elements align their baselines with the label instead of sitting above it
Fixed
compilerminimum lowered from0.14.2to0.14.0— patch releases add no new language features, so users on0.14.0/0.14.1were incorrectly excluded
[0.2.5] - 2026-01-27
Added
- Full manual (
docs/manual.typ+docs/manual.pdf) with structured examples - Expanded gallery with styled configuration and resume numbering examples
Fixed
- Indentation-independent parsing - Tasks now render identically regardless of source code indentation
- All
+items at any indentation level in the sametasks[]block are automatically flattened to the same level - Inconsistent indentation no longer creates nested enums with different numbering
- Moving one character in the source no longer breaks the layout
- All
- Column spanning support - Fixed
+(2)and+()syntax to work correctly with the new parser- Handles both
enum.itemnodes and text nodes starting with “+”
- Handles both
- Multiline tasks now capture unindented continuation lines as part of the previous item
Changed
- Internal parser completely rewritten to recursively extract and flatten all enum items
- Parent items with nested enums (from indentation) have enum nodes stripped from their content
- README gallery layout refreshed and feature list expanded
- Documentation updated for the new manual and examples
[0.2.0] - 2026-01-15
Fixed
- Indentation-independent parsing - Tasks now render identically regardless of source code indentation
- All
+items at any indentation level in the sametasks[]block are automatically flattened to the same level - Inconsistent indentation no longer creates nested enums with different numbering
- Moving one character in the source no longer breaks the layout
- All
- Column spanning support - Fixed
+(2)and+()syntax to work correctly with the new parser- Handles both
enum.itemnodes and text nodes starting with “+”
- Handles both
Changed
- Internal parser completely rewritten to recursively extract and flatten all enum items
- Parent items with nested enums (from indentation) have enum nodes stripped from their content
Added
- Column spanning - Items can span multiple columns using
(N)syntax (e.g.,+ (2) Contentor+ () Full width) label-baselineparameter for controlling vertical alignment of labels ("center","top","bottom", or custom length)label-weightparameter for bold labels ("regular"or"bold")- Advanced Features section in documentation with examples
- Gallery examples for column spanning and bold labels
Changed
- Repository URL updated to GitLab:
https://gitlab.com/nathan-ed/typst-package-taskize - Description enhanced to mention “Horizontal and vertical” flow directions
[0.1.0] - 2026-01-14
Added
- Initial release with horizontal columned list layout
- Flexible column counts (2, 3, 4, or custom)
- Multiple label formats (alphabetic, numeric, roman, bullets, custom)
- Resume numbering across multiple task blocks
- Two flow directions: horizontal (a b | c d) and vertical (a c | b d)
- Global configuration system with
tasks-setup() - Shorthand functions
tasks2,tasks3,tasks4 - Comprehensive parameter customization
- Full documentation with examples and parameter reference