Long-document structure, styles & templates
Apply heading styles, numbering, TOC, sections, headers/footers and templates to a spec.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- Word · Pages · Google Docs · LibreOffice Writer
- Class GUI-need label
G2· GUI-essential
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R04.02.01.01-T01 Apply heading styles and insert a TOC
Original instruction
In input/report.docx (40 pages, headings currently plain text), apply Heading 1/2/3 styles so the structure matches outline.txt, then insert an automatic Table of Contents on a new page after the title page. The TOC must list exactly the 23 reference headings.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Parsing the .docx: the 23 headings carry the correct w:pStyle (Heading1/2/3) matching outline.txt, and a TOC field exists listing exactly 23 entries with page numbers. Headings must use real styles; plain bold text at heading size is detected and fails. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% all headings styled at correct levels; 30% TOC field present; 30% 23 correct entries
R04.02.01.01-T02 Section breaks with per-section headers
Original instruction
In input/manual.docx, create three sections via section breaks and give each its own header text (Intro / Methods / Appendix) with page numbers continuing across sections. Headers must differ per section and not be linked to the previous.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolltype· Text & forms- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
The .docx has three sectPr with distinct header references (link-to-previous off) and the correct header text per section; page numbering is continuous. Graded on the section/header XML. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% three sections created; 30% distinct unlinked headers; 30% continuous page numbers
R04.02.01.01-T03 Insert a landscape section for a wide table
Original instruction
In input/data_report.docx, change only the section containing the wide table to landscape orientation (its own section), keeping all other pages portrait, and give the landscape section the header 'Appendix A'. Keep body content unchanged.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Exactly one sectPr is landscape (the table's section) with header 'Appendix A'; all other sections remain portrait; body text unchanged (hash of non-section-break content). Graded on section XML. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% table section isolated; 30% landscape only there; 30% header + rest portrait
R04.02.01.01-T04 Define a multilevel list scheme
Original instruction
In input/policy.docx, create and apply a multilevel list so chapter headings read '1.', sub-sections '1.1', and sub-sub '1.1.1', linked to Heading 1/2/3 styles. The numbering must auto-update and restart correctly per chapter.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
The numbering definition links levels to Heading1/2/3 and renders 1 / 1.1 / 1.1.1 with correct restart; numbers are field-driven (not typed). Graded on numbering XML + style links; typed literal numbers fail. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
30% numbering definition created; 40% linked to heading styles; 30% correct rendered sequence
R04.02.01.01-T05 Attach a template and map styles
Original instruction
Attach the template fixtures/corp_template.dotx to input/draft.docx and remap the document's existing custom styles to the template's Heading/Body/Caption styles per map.csv, so the document adopts the template's look without losing structure.
- Platforms
win· Windowsmac· macOS- Applications
- Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
The document references the template and paragraphs now carry the template's target styles per map.csv; the rendered fonts/spacing match the template's style definitions. Graded on the attached template + style assignments. Word VBA and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
30% template attached; 40% styles remapped per map.csv; 30% structure preserved
R04.02.01.01-T06 Restart appendix page numbering
Original instruction
In input/thesis.docx, start a new section at 'Appendix' and set its page numbering to the format A-1, A-2, ... while the main body keeps 1, 2, 3. Add a footer showing the page number in each section's format.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
The appendix sectPr sets pgNumType with the A-1 format and restart; the body keeps decimal numbering; footers show PAGE fields in the right format. Graded on section + field XML. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% appendix section; 30% A-1 format+restart; 30% body unchanged + footers
R04.02.01.01-T07 Create and apply a custom paragraph style
Original instruction
In input/notes.docx, create a paragraph style 'CalloutBox' (11pt, 6pt space-before/after, left indent 1cm, light-gray shading) and apply it to the five callout paragraphs listed in targets.txt, without using direct formatting.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
A style named CalloutBox exists with the specified properties and is applied (w:pStyle) to exactly the five target paragraphs; those paragraphs carry no overriding direct formatting. Graded on style definition + assignments. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% style defined to spec; 40% applied to the 5 paragraphs; 20% no direct-format override
R04.02.01.01-T08 Generate a TOC with exact entries
Original instruction
In input/handbook.docx the headings are already styled. Insert an automatic TOC that includes only Heading 1 and Heading 2 (not Heading 3), update it, and confirm it lists exactly the 14 reference entries with page numbers.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
A TOC field limited to levels 1-2 lists exactly 14 entries with correct page numbers; Heading 3 items are excluded. Graded on the TOC field + entry set. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% TOC field with level 1-2 scope; 30% 14 entries; 30% correct page numbers
R04.02.01.01-T09 Two-column layout for one section
Original instruction
In input/newsletter.docx, set only the middle 'Articles' section to a two-column layout with a 1cm gutter and a separating line, keeping the masthead and footer sections single-column. Use a section break, not text boxes.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
The Articles section's sectPr specifies two columns with the gutter and separator; adjacent sections remain single-column; no text boxes used. Graded on the column/section XML. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% section isolated; 40% two columns + gutter/line; 20% other sections single-column
R04.02.01.01-T10 Caption style and List of Figures
Original instruction
In input/figures.docx, add a styled caption 'Figure n: ...' beneath each of the 9 images using the Caption style and SEQ fields, then insert a List of Figures that lists all 9 with page numbers.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolltype· Text & forms- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Nine captions use the Caption style with SEQ Figure fields numbering 1-9; a List of Figures (TOF) field lists all 9 with page numbers. Graded on caption styles + SEQ/TOF fields; typed numbers fail. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% 9 styled captions with SEQ; 30% correct numbering; 30% List of Figures field
R04.02.01.01-T11 Convert manual bold headings to styles
Original instruction
In input/draft2.docx, the author used bold 14pt text instead of heading styles. Convert each such pseudo-heading to the matching Heading 1 or Heading 2 style (per the visual hierarchy) and remove the direct bold formatting.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
The former bold-text headings now carry Heading1/Heading2 styles matching the reference hierarchy, with the direct bold run property removed. Graded on w:pStyle assignments; remaining direct-format pseudo-headings fail. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
50% all pseudo-headings converted; 30% correct levels; 20% direct formatting removed
R04.02.01.01-T12 Cover page with different first-page header
Original instruction
In input/proposal.docx, add a cover page as its own section with a different first-page header (no page number on the cover), then start page 1 on the following page. Keep the rest of the document's headers intact.
- Platforms
win· Windowsmac· macOS- Applications
- Word · Pages
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
The cover is a separate section with titlePg (different first page) and no page number; numbering starts at 1 on the next page; other headers unchanged. Graded on section/header XML. Word VBA and COM/AppleScript automation (incl. Pages) are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% cover section + titlePg; 30% no number on cover; 30% numbering restart to 1
R04.02.01.01-T13 Spacing and widow/orphan control
Original instruction
In input/essay.docx, set Body Text style to 1.5 line spacing, 6pt space-after, and enable widow/orphan control and 'keep with next' on all Heading styles. Apply via the styles, not direct formatting.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scroll- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
The Body Text style definition shows 1.5 spacing + 6pt after; Heading styles have widowControl and keepNext set. Graded on the style definitions (not per-paragraph overrides). VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
50% body spacing in style; 50% widow/orphan + keep-with-next on headings
R04.02.01.01-T14 Fully linked numbered heading scheme
Original instruction
In input/standard.docx, build a complete linked numbering scheme where Heading 1-4 render as 1 / 1.1 / 1.1.1 / 1.1.1.1, numbers restart appropriately, cross-references to these numbers update, and the TOC shows the numbers. Must be style-driven and update on field refresh.
- Platforms
win· Windowsmac· macOS- Applications
- Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
The numbering definition links all four heading levels, renders the nested scheme with correct restarts, the TOC shows the numbers, and any heading-number cross-references resolve. Graded on numbering/style/TOC XML; typed numbers fail. Word VBA and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
30% 4-level numbering linked; 30% correct restarts/render; 20% TOC shows numbers; 20% cross-refs resolve
R04.02.01.01-T15 Apply a style set and font pairing
Original instruction
In input/brochure.docx, apply the built-in style set and the font pair specified in brief.txt (e.g. headings in one family, body in another) via the document's theme/style set so all styled text updates consistently. Do not change text content.
- Platforms
win· Windowsmac· macOS- Applications
- Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
The theme/style set defines the specified major/minor fonts and the styled headings/body resolve to them; text content unchanged. Graded on the theme font scheme + style definitions. Word VBA and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% style set applied; 40% correct font pairing in theme; 20% content unchanged
R04.02.01.01-T16 List of Tables index
Original instruction
In input/accounts.docx, caption the 6 tables as 'Table n' using the Caption style and SEQ Table fields, then insert a List of Tables (TOC of table captions) with page numbers after the main TOC.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolltype· Text & forms- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Six table captions use Caption style + SEQ Table (numbering 1-6) and a List of Tables field lists all 6 with page numbers. Graded on captions + TOF field. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% 6 styled table captions; 30% correct SEQ numbering; 30% List of Tables field
R04.02.01.01-T17 Structure a Google Doc with headings and TOC
Original instruction
In the Google Doc 'Team Handbook', apply Heading 1/2 styles to the section titles per outline.txt, insert a Table of Contents at the top, and confirm the document outline (left panel) lists all sections.
- Platforms
web· Any OS (web)- Applications
- Google Docs
- Interfaces
brow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
The Google Doc's named-style runs mark the correct headings matching outline.txt, a TOC element exists, and the document outline enumerates all sections. Graded on the doc's structural elements (styles, not bold text). The agent account has no Docs API or Apps Script access (OAuth scopes withheld), so the doc can be edited only in the browser editor; revision history must show the edits.
Proposed partial credit
40% headings styled; 30% TOC inserted; 30% outline complete
R04.02.01.01-T18 Author a reusable document template
Original instruction
Create a Word template corp_report.dotx with: a cover page, Heading 1-3 and Body/Caption styles to the spec in style_spec.csv, a header with a logo placeholder, automatic page numbers, and a TOC placeholder. Saving a new document from it must inherit all styles and layout.
- Platforms
win· Windowsmac· macOS- Applications
- Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolltype· Text & forms- Work product
WP01· Document- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
corp_report.dotx is a valid template whose styles match style_spec.csv, with the cover, header placeholder, page numbers, and TOC placeholder; a new doc created from it inherits them. Graded on the template's styles/layout + an instantiated doc. Word VBA and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
30% styles to spec; 25% cover+header placeholder; 20% page numbers+TOC placeholder; 25% inheritance works
R04.02.01.01-T19 Full long-document restructure to spec
Original instruction
In input/bigreport.docx (40 pages), implement the full spec in spec.md: apply Heading 1-3 styles per the outline, add front-matter (cover + auto TOC), three sections with distinct headers/footers, a landscape appendix with A-1 numbering, numbered headings, and a List of Figures. Content text must be preserved.
- Platforms
win· Windowsmac· macOS- Applications
- Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
The .docx satisfies each spec item: correct heading styles, TOC field, 3 sections with distinct headers, landscape A-1 appendix, linked heading numbers, and a List of Figures; body text hash matches the original content. Graded on the structural XML against the spec checklist. Word VBA and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
25% heading styles+TOC; 25% sections/headers; 25% appendix numbering+landscape; 25% numbered headings + LoF
R04.02.01.01-T20 Reconcile inconsistent styles to one set
Original instruction
input/merged.docx was assembled from several files and now has duplicate near-identical styles (Heading1, Heading 1, Heading1_copy, etc.). Consolidate every paragraph onto the single canonical style set in canon.csv and delete the redundant style definitions.
- Platforms
win· Windowsmac· macOS- Applications
- Word · LibreOffice Writer
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scrollvis· Visual judgment- Work product
WP01· Document- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Every paragraph maps to a canonical style from canon.csv, the redundant styles are removed from the style table, and the rendered appearance is unchanged. Graded on the style table + per-paragraph assignments. VBA, LibreOffice macros/UNO and COM/AppleScript automation are disabled, the VM has no terminal or Python, and writes to the file by any other process are logged and fail.
Proposed partial credit
40% paragraphs remapped to canonical styles; 30% redundant styles deleted; 30% appearance preserved
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.