Notes & knowledge-base organization
Capture, move, restructure, tag, link and dedupe notes and pages, including moving content between note and sticky-note apps.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- Notion · Obsidian · OneNote · Apple Notes
- 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.
R04.06.01.01-T01 File 30 loose Apple Notes into folders and pin two
Original instruction
In Apple Notes on the test Mac account, 30 notes sit in the default 'Notes' folder. Create folders 'Home', 'Work' and 'Travel', move each note into the folder given in input/note_folders.csv (matched by title), and pin 'Packing list' and 'Wi-Fi codes'. Delete nothing.
- Platforms
mac· macOS- Applications
- Apple Notes
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Grader reads NoteStore.sqlite after quitting Notes: folder membership per note equals CSV; the two notes pinned; note count unchanged (30). osascript/automation access to Notes is denied in the agent session.
Proposed partial credit
15% folders created; 65% notes filed (pro-rata over 30); 20% pins.
R04.06.01.01-T02 Add topic tags and aliases to notes in an Obsidian vault
Original instruction
In the Obsidian vault input/vault_research/, add YAML frontmatter to the 15 notes in folder 'Inbox': tags from input/topics.csv (lowercase, kebab-case) and an 'aliases' list with the abbreviations given there. Do not change note bodies.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
type· Text & formsclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Parse each note's frontmatter: tags and aliases equal CSV (order-insensitive); body text byte-identical; valid YAML. Obsidian metadata cache (grader-run) resolves aliases to the right notes.
Proposed partial credit
50% tags; 30% aliases; 20% bodies unchanged and YAML valid.
R04.06.01.01-T03 Create board and calendar views on a Notion task database
Original instruction
In the test Notion workspace, on database 'Team Tasks' add a Board view 'By status' grouped by Status (hide 'Archived' group), filtered to Assignee is not empty, sorted by Priority; and a Calendar view 'Due dates' by the Due property showing only open tasks.
- Platforms
web· Any OS (web)mac· macOSwin· Windows- Applications
- Notion
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Notion's public API cannot read or write views; grader's Playwright session opens each view's settings and reads layout, group-by, hidden groups, filters and sorts; compares with spec. Agent has no integration token.
Proposed partial credit
50% board view (group, hidden group, filter, sort); 40% calendar view (date prop, filter); 10% names.
R04.06.01.01-T04 Archive old sections and nest pages as subpages in OneNote
Original instruction
In OneNote notebook 'Lab Notebook' (test account), create section group 'Archive 2025', move the 6 sections with names ending '2025' into it, and in section 'Protocols' make the 12 pages listed in input/subpages.csv subpages (level 1 or 2) under their parent pages, in the listed order.
- Platforms
mac· macOSwin· Windows- Applications
- OneNote
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Graph API (read-only token held by grader): section group exists with exactly the 6 sections; page 'level' and 'order' in 'Protocols' equal CSV; no pages lost. No Graph credentials are issued to the agent and OneNote's Windows COM API is blocked; edits must come from the OneNote client.
Proposed partial credit
40% section group with right sections; 50% page levels and order; 10% nothing lost.
R04.06.01.01-T05 Tag Apple Notes and build a smart folder for 2026 receipts
Original instruction
In Apple Notes (test account), add the tag #receipt to the 14 notes listed in input/receipts.txt and #warranty to the 4 in input/warranty.txt, then create a Smart Folder 'Receipts 2026' that includes notes tagged #receipt and created in 2026.
- Platforms
mac· macOSios· iOS/iPadOS- Applications
- Apple Notes
- Interfaces
desk· Desktop GUImob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
type· Text & formsclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
NoteStore.sqlite: tag associations equal lists; a smart folder with query tags=receipt AND creation date in 2026; its computed membership equals key (11 notes). osascript and Shortcuts access to Notes is disabled for the agent, and NoteStore.sqlite is read-only to it.
Proposed partial credit
40% #receipt tags; 15% #warranty; 45% smart folder query and membership.
R04.06.01.01-T06 Merge duplicate notes in Obsidian and keep every backlink working
Original instruction
input/vault_pm/ has 9 pairs of duplicate notes (same topic, different titles). For each pair, merge the shorter into the longer (append unique paragraphs), rename the survivor to the canonical title in input/canonical.csv, and make every existing link to either note resolve to the survivor.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsnav· Pan/zoom/scroll- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Vault: 9 survivors with canonical filenames, losers deleted; survivor bodies contain all unique paragraphs of both (set comparison); zero unresolved links in Obsidian metadata cache; link count per survivor equals union of prior backlinks.
Proposed partial credit
30% merges complete; 20% canonical names; 40% backlinks resolved; 10% no lost content.
R04.06.01.01-T07 Relate Notion projects to tasks with a rollup and per-project views
Original instruction
In the test workspace, add a relation between databases 'Projects' and 'Tasks', link each task to its project per input/task_project.csv, add a rollup '% done' on Projects (percent of related tasks with Status = Done), and on each of the 5 project pages insert a linked view of Tasks filtered to that project.
- Platforms
web· Any OS (web)mac· macOSwin· Windows- Applications
- Notion
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader's read-only integration: relation values equal CSV; rollup values equal key per project; Playwright inspection of each project page finds a linked database view with filter Project contains <this page>. Linked views are not creatable via API. Agent has no integration token.
Proposed partial credit
35% relations set; 25% rollup correct; 40% five filtered linked views.
R04.06.01.01-T08 Tag action items and decisions across OneNote meeting pages
Original instruction
In OneNote section 'Weekly syncs' (8 pages), apply the To Do tag to every line starting 'AI:' and the Important tag to every line starting 'DECISION:', remove those prefixes, then create a tag summary page listing all To Do items grouped by page.
- Platforms
mac· macOSwin· Windows- Applications
- OneNote
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickkey· Shortcuts- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Graph page HTML: data-tag='to-do' on the 23 action lines and 'important' on the 9 decision lines, prefixes removed; a summary page containing all 23 items grouped under page titles. No Graph credentials are issued to the agent and OneNote's Windows COM API is blocked; edits must come from the OneNote client (page lastModified by client).
Proposed partial credit
40% to-do tags; 20% important tags; 15% prefixes removed; 25% summary page.
R04.06.01.01-T09 Lay out project notes on an Obsidian canvas by status with dependencies
Original instruction
In vault input/vault_roadmap/, create canvas 'Roadmap.canvas' with the 12 notes in folder 'Epics' as file cards arranged in 3 columns (Todo, Doing, Done per each note's status field), cards in a column aligned and evenly spaced, and directed edges for the 9 dependencies in input/deps.csv.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropgeom· Precise geometryclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Parse .canvas JSON: 12 file nodes; x ranges cluster into 3 columns by status; same-column nodes share x (±10) with equal y gaps (±10); 9 edges with fromNode/toNode matching deps direction; 3 column header text nodes. Agent session has no shell or file-editing tools and Obsidian's developer console and community plugins are disabled; the canvas must be produced in Obsidian.
Proposed partial credit
25% cards present; 35% column placement; 15% alignment/spacing; 25% edges.
R04.06.01.01-T10 Move sixty loose Notion pages into a database with typed properties
Original instruction
In the test workspace, 60 meeting-note pages sit under 'Old notes'. Create database 'Meetings' (Date, Type select, Owner person, Project relation), move every page into it, fill properties from each page's first lines, and add a template 'Weekly sync' with a standard agenda.
- Platforms
web· Any OS (web)mac· macOSwin· Windows- Applications
- Notion
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropclick· Point & clicktype· Text & forms- Work product
WP08· App/system state- Conditions
long· Long-horizon- Proposed difficulty
- 4 / 5
Proposed verifier
Read-only API: 60 pages now parented by 'Meetings' (page ids preserved, not copies); property values equal key; template existence checked via Playwright (templates not exposed by API). Agent has no integration token, so pages must be moved in the Notion UI.
Proposed partial credit
35% pages moved (ids preserved); 40% properties correct; 10% schema; 15% template.
R04.06.01.01-T11 Fix broken wikilinks after a vault restructure and build topic MOCs
Original instruction
input/vault_course/ was reorganized into new folders and renamed, leaving 23 broken wikilinks. Repair every link to point at the right note (use input/rename_map.csv), then create one MOC note per tag listed in input/mocs.txt linking all notes with that tag, sorted alphabetically.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
type· Text & formsclick· Point & clicknav· Pan/zoom/scroll- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Obsidian metadata cache (grader-run headless): 0 unresolved links; each fixed link targets the mapped note; each MOC lists exactly the tagged notes in alphabetical order; no other notes edited beyond link text.
Proposed partial credit
50% links repaired (pro-rata over 23); 40% MOCs complete and sorted; 10% no collateral edits.
R04.06.01.01-T12 Consolidate duplicate recipe notes with checklists and note links
Original instruction
In Apple Notes folder 'Recipes', 8 duplicate notes cover 4 recipes. Merge each pair into one note (keep all unique steps), turn ingredient lists into checklists, add a note link ('>>' link) from each recipe to its 'Goes well with' recipe per input/pairings.csv, and delete the duplicates.
- Platforms
mac· macOSios· iOS/iPadOS- Applications
- Apple Notes
- Interfaces
desk· Desktop GUImob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
type· Text & formsclick· Point & click- Work product
WP09· Files & folders- Conditions
risk· Irreversible/risky- Proposed difficulty
- 4 / 5
Proposed verifier
NoteStore: 4 notes remain in 'Recipes' (duplicates in Recently Deleted, recoverable); each contains all unique step lines; ingredient lines are checklist items; note-link attachments point to the right note ids. osascript and Shortcuts access to Notes is disabled for the agent.
Proposed partial credit
35% merges complete; 25% checklists; 25% note links; 15% duplicates deleted only after merge.
R04.06.01.01-T13 Split a 40-page OneNote section into four topic sections
Original instruction
OneNote section 'Misc' in notebook 'Ops Handbook' (shared via a test SharePoint site) holds 40 pages on 4 topics. Create sections 'Hardware', 'Accounts', 'Vendors', 'Procedures', move each page to its section per input/page_topics.csv preserving subpage nesting, then delete the empty 'Misc' section.
- Platforms
mac· macOSwin· Windows- Applications
- OneNote
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropclick· Point & click- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Graph read: 4 sections with page sets equal CSV; subpage parent/level relations preserved; page ids preserved (moved, not copied); 'Misc' absent; page count 40. No Graph credentials are issued to the agent and OneNote's Windows COM API is blocked; edits must come from the OneNote client (page lastModified by client).
Proposed partial credit
15% sections; 55% pages moved correctly; 20% nesting preserved; 10% Misc deleted only after.
R04.06.01.01-T14 Build a department wiki in Notion with synced headers and SOP views
Original instruction
In the test workspace build 'Ops Wiki' per input/wiki_spec.md: 4 department pages under it, a synced block (contacts banner) on all 4, an SOP database with Department and Review Date, one filtered linked view per department page, and move the 18 existing SOP pages into the database.
- Platforms
web· Any OS (web)mac· macOSwin· Windows- Applications
- Notion
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickdrag· Drag & droptype· Text & forms- Work product
WP08· App/system state- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
API read: page tree per spec; the 4 department pages contain synced_block children referencing one original; 18 SOP pages parented by the database with properties set; Playwright reads each linked view's filter = that department. Agent has no Notion integration token.
Proposed partial credit
20% page tree; 20% synced blocks; 30% SOPs moved with properties; 30% filtered views.
R04.06.01.01-T15 Atomize mega-notes and normalize metadata in a 300-note vault
Original instruction
In input/vault_zk/ (300 notes): normalize frontmatter (dates ISO 8601, tags lowercase kebab, 'type' field required), split the 5 notes over 3,000 words into atomic notes at H2 boundaries with links back to a stub parent, update all inbound links to point at the right atomic note, and add a Dataview index note.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
type· Text & formsclick· Point & clickcode· Code & commands- Work product
WP09· Files & folders- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Frontmatter schema check on all notes; 5 parents become stubs linking all children; child notes equal H2 sections (text match); inbound heading links (note#Heading) retargeted to child notes; 0 unresolved links; Dataview index query renders expected rows (headless).
Proposed partial credit
30% frontmatter normalized; 30% splits; 25% links retargeted; 15% Dataview index.
R04.06.01.01-T16 Migrate a OneNote section into Notion keeping structure and to-dos
Original instruction
Move OneNote section 'Kitchen remodel' (25 pages with images, tables and To Do tags) into a Notion page 'Kitchen remodel' in the test workspace: one subpage per OneNote page in the same order, images inline, tables as Notion tables, To Do tags as checkbox blocks with the same checked state.
- Platforms
mac· macOSwin· Windowsweb· Any OS (web)- Applications
- OneNote · Notion
- Interfaces
desk· Desktop GUIbrow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
multi· Cross-appclick· Point & clickdrag· Drag & drop- Work product
WP09· Files & folders- Conditions
multiapp· Multi-applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Notion API read: 25 child pages in order with titles; per page text similarity ≥0.95 vs OneNote HTML; image count/hashes match; table blocks with equal cells; to_do blocks with matching checked state. Agent has no Notion or Graph token and OneNote's Windows COM API is blocked; content must be moved through the apps.
Proposed partial credit
20% pages and order; 25% text; 35% images and tables; 20% to-dos.
R04.06.01.01-T17 Control pair of T11: repair the broken wikilinks with a script
Original instruction
Control pair of T11. From the terminal, write a script that rewrites the 23 broken wikilinks in input/vault_course/ using input/rename_map.csv (handling aliases and heading anchors) and reports any it cannot resolve.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Obsidian · 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
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Same link checker as T11 (0 unresolved, correct targets, no collateral edits); compared against the GUI variant.
Proposed partial credit
80% links repaired (pro-rata); 20% no collateral edits.
R04.06.01.01-T18 Reorganize OneNote pages and tags by touch on an Android phone
Original instruction
In OneNote for Android (test account), in notebook 'Garden', move the 7 pages listed in input/moves.csv to their target sections by long-press and move, make 'Seed list' a subpage of 'Spring plan', and add the To Do tag to the 5 lines starting 'Buy'.
- Platforms
and· Android- Applications
- OneNote
- Interfaces
mob· Mobile touch- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Graph read after sync: page section membership equals CSV; 'Seed list' level 1 under 'Spring plan'; data-tag to-do on the 5 lines; page count unchanged. No Graph credentials are issued to the agent and adb shell is disabled; edits must come from the OneNote app (page lastModified by client).
Proposed partial credit
50% page moves; 20% subpage; 30% to-do tags.
R04.06.01.01-T19 Turn a pasted list into a sorted checklist in Notes on iPhone
Original instruction
On iPhone (test Apple ID), in note 'Move-out tasks' convert the 16 pasted lines into a checklist, drag items into the room order given in the note's first line, check off the 4 tasks already done per input/done.txt, enable 'Sort Checked Items' to bottom, and pin the note.
- Platforms
ios· iOS/iPadOS- Applications
- Apple Notes
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP09· Files & folders- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
NoteStore (synced to grader Mac): 16 checklist items in the specified order (unchecked part), 4 checked items done; note pinned; checked-sort setting on. Shortcuts access to Notes is disabled on the agent's iPhone, so the checklist is edited in Notes itself.
Proposed partial credit
25% checklist conversion; 35% order; 20% checked items; 20% sort setting + pin.
R04.06.01.01-T20 De-duplicate a Notion database and merge inconsistent select options
Original instruction
Database 'Leads' in the test workspace has 40 duplicate entries (same Email) and Status options 'In progress', 'In Progress' and 'WIP'. Merge duplicates into the oldest entry (union of non-empty properties, newest Notes appended), collapse Status to 'In progress', and archive the duplicates.
- Platforms
web· Any OS (web)mac· macOSwin· Windows- Applications
- Notion
- Interfaces
brow· Browserdesk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP08· App/system state- Conditions
risk· Irreversible/risky- Proposed difficulty
- 4 / 5
Proposed verifier
API read: one active entry per Email; survivors are the oldest ids with merged property values equal key; Status options set equals spec and all entries mapped; duplicates archived (recoverable), not hard-deleted.
Proposed partial credit
40% duplicates merged into right survivor; 25% property union; 20% status options; 15% archived not deleted.
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.