Computer-Use Agent Taxonomy中文
R06.01.01.03 · TASK CLASS

Brand template adaptation & resizing

Adapt a brand template to new copy and sizes without breaking rules.

20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.

Typical applications
Canva · Figma · Illustrator
Class GUI-need label
HY · Hybrid

Task proposals · 20

Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.

R06.01.01.03-T01 Adapt a square social post to three formats within brand rules

Original instruction

In Canva (test team with Brand Kit), adapt design 'Fall Launch 1080x1080' into a 1080x1920 story, a 1200x628 link ad and an A4 flyer. Follow input/brand_rules.pdf: logo clear space >= 0.5x logo height, all text inside each format's safe zone (story: 250 px top/bottom), brand palette only, no image distortion. Export each as PDF to output/.

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
drag · Drag & drop geom · Precise geometry vis · Visual judgment
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Exported PDFs parsed: page sizes exact; text bboxes inside safe zones; logo clear-space box free of other objects; fill colors within DeltaE2000 <= 1 of palette; image aspect ratios preserved. No Canva API key issued.

Proposed partial credit

15% sizes; 30% safe zones; 25% logo clear space; 30% palette and no distortion.

R06.01.01.03-T02 Generate certificates from a list with no text overflow

Original instruction

In Canva (test team), use Bulk Create on template 'Workshop Certificate' with input/attendees.csv (25 names, some long, dates, course titles) to produce 25 pages. Every name must fit one line inside its frame at >= 24 pt (shrink only names that need it). Export one PDF to output/certificates.pdf.

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click type · Text & forms vis · Visual judgment
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

PDF: 25 pages; per page the name/date/course text equals CSV row; name on one line within frame bbox, size >= 24 pt; other elements unchanged from template.

Proposed partial credit

35% correct data per page; 40% names fit one line; 15% min size; 10% template intact.

R06.01.01.03-T03 Restyle a data chart into the brand template proportionally

Original instruction

In Illustrator, rebuild the bar chart from input/q3_data.csv inside the brand template input/brand_chart.ait: bars in brand blue, heights proportional to values within 1%, baseline at y=0, category labels in the brand font 9 pt, value labels above bars. Save output/q3_chart.ai and export output/q3_chart.svg.

Platforms
mac · macOS win · Windows
Applications
Illustrator
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
geom · Precise geometry type · Text & forms click · Point & click
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
3 / 5

Proposed verifier

SVG: bar rect heights / values proportional (max deviation <= 1%); bars share baseline; fill equals brand hex; text nodes/outlines match labels and font (from .ai). ExtendScript disabled.

Proposed partial credit

45% proportional heights; 15% baseline; 20% labels; 20% brand styling.

R06.01.01.03-T04 Lay out a business card with bleed and a placed QR code

Original instruction

In Affinity Designer or Illustrator, build a 85x55 mm business card from input/brand_kit/ and input/card_grid.pdf: name, title, contacts on the grid, logo top-left, a QR code encoding the vCard in input/vcard.vcf at >= 15 mm placed on the grid, background extended to a 3 mm bleed. Export output/card.pdf (PDF/X-4).

Platforms
mac · macOS win · Windows
Applications
Affinity Designer · Illustrator
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
geom · Precise geometry drag · Drag & drop type · Text & forms
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
3 / 5

Proposed verifier

pikepdf: TrimBox 85x55 mm, BleedBox +3 mm, PDF/X-4 OutputIntent; QR decodes to the vCard and is >= 15 mm; text/logo positions within 0.5 mm of grid; background reaches bleed.

Proposed partial credit

25% boxes and bleed; 25% QR content and size; 35% grid placement; 15% PDF/X.

R06.01.01.03-T05 Make a resizable social template with auto layout and constraints

Original instruction

In Figma (test file), turn frame 'Promo 1080' into a component with auto layout and constraints so that resizing to 1080x1350, 1080x1920 and 1200x628 keeps the logo pinned top-left with 64 px margin, headline wrapping inside 80% width, and CTA pinned bottom. Create the 3 resized instances on a page 'Sizes'.

Platforms
mac · macOS win · Windows web · Any OS (web)
Applications
Figma
Interfaces
desk · Desktop GUI brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
drag · Drag & drop geom · Precise geometry click · Point & click
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Figma REST: component with layoutMode/constraints; 3 instances with exact sizes; in each, logo at (64,64), headline width <= 0.8 frame width, CTA bottom margin per spec, no clipped text. Plugins disabled.

Proposed partial credit

30% component setup; 20% instance sizes; 35% layout rules hold in all sizes; 15% no clipping.

R06.01.01.03-T06 Apply a brand kit to a template and resize it

Original instruction

In Adobe Express (test account with Brand), open template 'Webinar Promo', apply the brand fonts and colors from the 'Northwind' brand, replace the copy with input/copy.txt, and resize to Instagram post, LinkedIn post and YouTube thumbnail. Download all three as PNG to output/.

Platforms
web · Any OS (web)
Applications
Adobe Express
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click type · Text & forms
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

PNGs: sizes exact; OCR text equals copy; dominant text/fill colors within DeltaE2000 <= 2 of brand; font identified via glyph template match.

Proposed partial credit

25% sizes; 30% copy; 25% colors; 20% fonts.

R06.01.01.03-T07 Replace copy in a slide template without overflow

Original instruction

In Canva (test team), duplicate presentation template 'Brand Deck' and replace the copy on its 6 slides with input/deck_copy.md, keeping brand fonts and sizes; where text is longer, edit layout boxes rather than shrinking below the template's minimum 18 pt. Export output/deck.pdf.

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
type · Text & forms drag · Drag & drop
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

PDF: text per slide equals copy; no text outside its frame or slide; min size >= 18 pt; fonts match brand.

Proposed partial credit

40% copy; 35% no overflow; 15% min size; 10% fonts.

R06.01.01.03-T08 Localize a banner to German within the safe area

Original instruction

In Illustrator, localize input/banner_en.ai (728x90 and 300x250 artboards) to German using input/de_copy.txt. German copy is longer: fit it inside each safe area by stepping down the brand type scale (only sizes in input/type_scale.txt), never using horizontal scaling or condensed fonts. Export SVG and PNG per artboard.

Platforms
mac · macOS win · Windows
Applications
Illustrator
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
type · Text & forms vis · Visual judgment geom · Precise geometry
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
3 / 5

Proposed verifier

Per artboard: text equals German copy; text bboxes inside safe area; font sizes in allowed set; horizontal scale 100% and font family unchanged.

Proposed partial credit

30% copy; 40% inside safe area; 20% allowed sizes; 10% no scaling.

R06.01.01.03-T09 Rebrand a product file by swapping to new variables

Original instruction

In Figma (test file 'App Screens', 20 frames), migrate from old brand to new: replace every hard-coded fill/stroke color and text style with the new brand's variables and styles from the published library 'Brand 2027' using input/token_map.csv. When done, no node may use a raw color.

Platforms
mac · macOS win · Windows web · Any OS (web)
Applications
Figma
Interfaces
desk · Desktop GUI brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click nav · Pan/zoom/scroll multi · Cross-app
Work product
WP03 · Visual layout
Conditions
long · Long-horizon
Proposed difficulty
4 / 5

Proposed verifier

Figma REST: all fills/strokes bound to variables (boundVariables) mapped per CSV; text nodes use new styles; 0 raw colors; visual diff limited to color changes (layout bbox unchanged).

Proposed partial credit

45% colors bound correctly; 30% text styles; 15% zero raw colors; 10% layout unchanged.

R06.01.01.03-T10 Audit and fix brand violations across ten designs

Original instruction

input/designs/ holds 10 Canva designs (shared to the test team) made by regional teams. Using input/brand_rules.pdf, find and fix violations: off-palette colors, stretched or recolored logos, logo too close to edges, non-brand fonts, low-res images under 150 dpi at print size. Log each fix in output/audit.csv (design, issue, fix).

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
vis · Visual judgment click · Point & click drag · Drag & drop
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Exports after fix: palette, logo aspect and clear space, fonts, image effective dpi all compliant; audit.csv recall of the 23 seeded violations; no new violations introduced. No Canva Connect API key is issued and apps are disabled for the test team, so fixes must be made in the Canva editor.

Proposed partial credit

50% violations fixed; 30% audit recall; 20% no regressions.

R06.01.01.03-T11 Resize an email header and ad set then fix the auto-resize

Original instruction

In Canva (test team), use Resize on 'Holiday Hero' to create a 600x200 email header and 300x600, 160x600 and 970x250 display ads. Auto-resize leaves overlaps and cropped text: fix each so the headline, CTA and logo are visible, unoverlapped and within 10 px margins. Export PNGs to output/.

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
drag · Drag & drop vis · Visual judgment geom · Precise geometry
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

PNG + design export (PDF): sizes exact; headline/CTA/logo bboxes present, non-overlapping, inside 10 px margin; OCR text complete.

Proposed partial credit

15% sizes; 35% no overlaps; 30% margins; 20% text complete.

R06.01.01.03-T12 Data-merge forty name badges with photos

Original instruction

In InDesign, use Data Merge with input/attendees.csv (40 rows, name, company, role, photo path) on template input/badge.indd (4 per A4 page) so photos fit frames proportionally and centered, names over 18 characters drop to the 'Name Long' style, and role colors follow the role. Export output/badges.pdf.

Platforms
mac · macOS win · Windows
Applications
InDesign
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
click · Point & click type · Text & forms
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
3 / 5

Proposed verifier

PDF: 10 pages, 40 badges with correct text; image placements proportional (no distortion) and centered; long names use Name Long size; role color per mapping.

Proposed partial credit

35% text merged; 25% photo fitting; 20% long-name style; 20% role colors.

R06.01.01.03-T13 Adapt a poster grid to three paper sizes

Original instruction

In Affinity Designer, adapt input/poster_A2.afdesign to A3, A1 and US Tabloid. Keep the 12-column grid proportional to each format's width, preserve the type hierarchy ratios, keep 3 mm bleed, and reflow (not stretch) the photo crop so the subject's face stays in the upper third. Export PDF/X-4 per size.

Platforms
mac · macOS win · Windows
Applications
Affinity Designer
Interfaces
desk · Desktop GUI
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G3 · GUI-only
Interaction primitives
drag · Drag & drop geom · Precise geometry vis · Visual judgment
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
3 / 5

Proposed verifier

PDF boxes correct per size; element x-positions align to scaled grid columns (+/-0.5 mm); type size ratios equal original; photo not distorted; face bbox in upper third. The VM has no shell or scripting runtime and the PDF Creator field must name Affinity, so files not exported by the app are detected.

Proposed partial credit

20% sizes and bleed; 35% grid alignment; 20% type ratios; 25% photo crop.

R06.01.01.03-T14 Control: generate the T02 certificates by script

Original instruction

Control pair of T02. From input/attendees.csv and the exported blank template input/certificate_bg.pdf, write a script (e.g. Python + reportlab/pypdf) producing output/certificates.pdf with the same text placement rules (one-line names >= 24 pt, auto-shrink only when needed).

Platforms
lin · Linux mac · macOS win · Windows
Applications
Python
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

Same PDF checks as T02.

Proposed partial credit

35% data; 40% fit; 15% min size; 10% template intact.

R06.01.01.03-T15 Adapt a key visual to a billboard with legibility minimums

Original instruction

In Illustrator, adapt key visual input/kv_master.ai to a 48-sheet billboard (6096x3048 mm at 1:10 scale artboard) following input/ooh_rules.pdf: at most 7 words, headline cap height >= 1/10 of board height, logo >= 15% of width, no small legal text except the 1 required line at minimum size. Export output/billboard.pdf.

Platforms
mac · macOS win · Windows
Applications
Illustrator
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
geom · Precise geometry vis · Visual judgment type · Text & forms
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app
Proposed difficulty
4 / 5

Proposed verifier

PDF: artboard size; word count <= 7 excluding legal; measured cap height and logo width ratios meet rules; legal line present at min size; contrast ratio headline/background >= 7:1.

Proposed partial credit

20% size; 30% headline legibility; 20% logo size; 30% legal line and contrast.

R06.01.01.03-T16 Build a seamless five-slide carousel

Original instruction

In Figma (test file), build a 5-slide 1080x1350 Instagram carousel from input/carousel_brief.md where the background illustration and a product photo cross slide boundaries continuously, using one wide frame sliced into 5 frames. Export the 5 PNGs; adjacent edges must line up pixel-exactly.

Platforms
mac · macOS win · Windows web · Any OS (web)
Applications
Figma
Interfaces
desk · Desktop GUI brow · Browser
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
drag · Drag & drop geom · Precise geometry vis · Visual judgment
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

PNGs: 5 at 1080x1350; adjacent edge columns match (mean abs diff <= 1 level); crossing elements continuous; brief copy per slide via OCR; safe margins respected. Figma plugins and REST API tokens are withheld from the test account, so the carousel must be built and exported in the editor.

Proposed partial credit

40% seamless edges; 20% sizes; 25% copy per slide; 15% margins.

R06.01.01.03-T17 Fix contrast in a template within the brand palette

Original instruction

In Canva (test team), template 'Event Promo' has 6 text elements failing WCAG 2.1 contrast. Change only colors (text or its background shape) using brand palette colors so every text reaches >= 4.5:1 (>= 3:1 for text >= 24 pt bold), without changing layout. Export output/event_promo.pdf.

Platforms
web · Any OS (web)
Applications
Canva
Interfaces
brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click vis · Visual judgment
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

PDF: computed contrast per text element meets thresholds; all colors in palette; geometry identical to original (bbox diff 0).

Proposed partial credit

60% contrast thresholds; 25% palette compliance; 15% layout unchanged.

R06.01.01.03-T18 Produce a full campaign kit from a master key visual

Original instruction

From input/kv_master.ai and input/deliverables.xlsx (12 formats: social, display, print, email, OOH), produce all deliverables in Illustrator with brand rules (input/brand_rules.pdf): safe zones, logo clear space, legal line on print/OOH, file specs (PNG <= 150 KB for display, PDF/X-4 for print). Save to output/kit/ with exact file names.

Platforms
mac · macOS win · Windows
Applications
Illustrator
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
geom · Precise geometry drag · Drag & drop vis · Visual judgment multi · Cross-app
Work product
WP03 · Visual layout
Conditions
licensed · Licensed pro app long · Long-horizon
Proposed difficulty
5 / 5

Proposed verifier

Per file: name, dimensions, format and size limits; safe zones and clear space via parsed geometry; legal line presence where required; palette compliance; total 12 files.

Proposed partial credit

20% file specs; 30% safe zones; 20% clear space; 30% legal lines and palette.

R06.01.01.03-T19 Build a location template system and export 30 variants

Original instruction

In Figma (test file), build a store-opening poster template whose city name, address, date, and map image come from variables with 30 modes (input/stores.csv), with auto layout so long city names wrap without overlapping the map. Create 30 variant frames bound to each mode and export 30 PNGs named by store id.

Platforms
mac · macOS win · Windows web · Any OS (web)
Applications
Figma
Interfaces
desk · Desktop GUI brow · Browser
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click type · Text & forms multi · Cross-app
Work product
WP03 · Visual layout
Conditions
long · Long-horizon
Proposed difficulty
5 / 5

Proposed verifier

Figma REST: variable collection with 30 modes and values per CSV; 30 frames with explicit modes; PNGs OCR to correct text; no overlaps between text and map bboxes.

Proposed partial credit

30% variables/modes; 25% frames bound; 25% exports correct; 20% no overlaps.

R06.01.01.03-T20 Recrop a story photo on a phone to keep the subject safe

Original instruction

On an iOS simulator with the Canva app (test account), open 'Story Template - Coach', replace the photo with input/coach.jpg from the photo library, then pan and zoom the crop by touch so the coach's face sits inside the safe zone (y 250-1670) and in the upper half, without upscaling beyond 150%. Download the PNG.

Platforms
ios · iOS/iPadOS
Applications
Canva
Interfaces
mob · Mobile touch
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G3 · GUI-only
Interaction primitives
touch · Touch gestures vis · Visual judgment geom · Precise geometry
Work product
WP03 · Visual layout
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

PNG 1080x1920; face detection bbox inside safe zone and upper half; effective scale <= 150% (image crop metadata from design export); other layers unchanged. Canva Connect API keys are withheld and the simulator accepts only touch input, so the crop must be set in the app.

Proposed partial credit

45% face placement; 25% scale limit; 20% photo replaced; 10% other layers intact.

Mapped research references

Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.

In this workflow family