Flowcharts, BPMN & org charts
Build and edit connected diagrams that follow notation 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
- Visio · draw.io · Lucidchart · OmniGraffle
- Class GUI-need label
G3· GUI-only
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R11.03.01.01-T01 Insert a cache node between API and DB and reroute the arrows
Original instruction
Open diagrams/arch.drawio in draw.io desktop. Insert a 'Cache' rectangle between the 'API' and 'DB' shapes, reroute so the only edges are API→Cache and Cache→DB (delete the old API→DB edge), with both edges glued to shape connection points. Save.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- draw.io
- 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
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Parse .drawio XML: vertex labeled Cache exists between API and DB (x-center between them); edges have source/target ids API→Cache, Cache→DB; no API→DB edge; no dangling edge (both ends attached). mxfile host/agent attributes and compressed diagram show it was saved by draw.io, not hand-edited.
Proposed partial credit
25% Cache node; 50% edge set exact; 25% edges glued
R11.03.01.01-T02 Recreate a flowchart image as a connected Visio diagram
Original instruction
In Visio, recreate the flowchart in input/flow.png (5 process boxes, 1 decision diamond, 6 arrows incl. Yes/No branches) on a Basic Flowchart page, with connectors glued to shapes and branch labels. Save as output/flow.vsdx.
- Platforms
win· Windowsweb· Any OS (web)- Applications
- Visio
- Interfaces
desk· Desktop GUIbrow· Browser- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & droptype· Text & formsgeom· Precise geometry- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Parse VSDX: shape masters (Process/Decision) and text; connector Connects table shows both ends glued; directed graph isomorphic to reference with labels equal; Yes/No labels on decision outgoing edges. Compared with control pair T17. Visio VBA/COM automation is disabled, no Graph credentials are issued and the agent has no terminal or Python; the .vsdx must be saved by Visio.
Proposed partial credit
25% shapes and text; 40% graph isomorphic; 20% glued connectors; 15% branch labels
R11.03.01.01-T03 Draw a cross-functional purchase approval flow from an SOP
Original instruction
From sop_purchase.docx, build a Visio cross-functional flowchart with 4 swimlanes (Requester, Manager, Procurement, Finance), 12 steps and 3 decisions in the lane responsible for each, connectors glued, and phases 'Request' and 'Fulfil'. Save output/purchase_cff.vsdx.
- Platforms
win· Windows- Applications
- Visio · Word
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & droptype· Text & formsgeom· Precise geometry- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro appmultiapp· Multi-app- Proposed difficulty
- 4 / 5
Proposed verifier
Parse VSDX: 4 lane containers with names; each step/decision shape contained in the gold lane (container membership); graph matches gold (edit distance ≤2); phase separators exist; all connectors glued. Visio VBA and COM automation are disabled by policy and the agent has no terminal or Python, so the .vsdx must be saved by Visio (docProps/app.xml checked).
Proposed partial credit
20% lanes; 35% steps in correct lanes; 30% flow matches; 15% phases and glue
R11.03.01.01-T04 Model a BPMN process with a timer escalation and deploy it
Original instruction
In Camunda Modeler, model process 'Leave Request' from leave_process.md: start event, user tasks Submit/Review, exclusive gateway approved?, a 48-hour non-interrupting timer on Review that notifies HR, end events Approved/Rejected. Make it pass bpmnlint 'recommended' and deploy to the local test engine.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Camunda Modeler
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropclick· Point & clicktype· Text & forms- Work product
WP06· Code/engineering- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
BPMN XML: elements and types match spec; timer boundary cancelActivity=false with PT48H; gateway with two conditional flows; bpmnlint recommended passes; diagram DI has shapes for all elements. Engine REST shows deployment; test instance completes both paths. Agent has no terminal, text editor or HTTP client, so the BPMN must be saved by Camunda Modeler (exporter attribute checked) and deployed from its Deploy dialog.
Proposed partial credit
35% elements; 20% timer boundary; 15% lint clean; 30% deployed and runs
R11.03.01.01-T05 Import an org chart from HR data and fix reporting lines
Original instruction
In Lucidchart (test account), generate an org chart by importing hr_roster.csv (name, title, manager id, dept). Three managers changed last week (changes.txt); fix those reporting lines by dragging, then apply the tree layout and export output/org.pdf.
- Platforms
web· Any OS (web)- Applications
- Lucidchart
- Interfaces
brow· Browser- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropclick· Point & click- Work product
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Lucid document export (verifier API or .vsdx export): parent-child edges equal updated roster tree (68 nodes); titles present; layout is tree (each child below parent). org.pdf text contains all names. No Lucid API key or OAuth app is issued to the agent, so edits must come from Lucidchart browser sessions (checked in document revision history).
Proposed partial credit
30% import complete; 40% three changes applied with tree consistent; 15% layout; 15% PDF export
R11.03.01.01-T06 Build a color-coded department org chart in OmniGraffle
Original instruction
In OmniGraffle, build an org chart for the 25 people in dept_eng.csv using hierarchical auto-layout (top-to-bottom), each box showing name and title, fill color by office (Berlin blue #2F6DB5, Austin orange #E07A1F, Remote grey #8A8A8A). Save output/eng_org.graffle and export PDF.
- Platforms
mac· macOS- Applications
- OmniGraffle
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & droptype· Text & formsclick· Point & click- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Parse .graffle (zipped plist): 25 shapes with text; lines connect manager→report matching CSV tree; fill colors per office ΔE≤2; vertical levels consistent with depth. PDF exported. Omni Automation and AppleScript access to OmniGraffle are disabled (TCC denies Automation) and the agent has no terminal; the file must be saved by OmniGraffle.
Proposed partial credit
25% boxes and text; 40% hierarchy; 20% colors; 15% export
R11.03.01.01-T07 Connect services to a gateway and align them in one row
Original instruction
In draw.io open diagrams/services.drawio. Connect every shape whose label starts with 'Service' to the 'Gateway' shape with an arrow from Gateway to the service, then align all service shapes in one horizontal row with equal spacing. Save.
- Platforms
mac· macOSwin· Windowslin· Linuxweb· Any OS (web)- Applications
- draw.io
- Interfaces
desk· Desktop GUIbrow· Browser- 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
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Parse XML: one edge Gateway→each of the 7 services (glued), no duplicates; service y-centres equal ±5 px; horizontal gaps equal ±5 px; no other shapes moved. Agent has no terminal or text editor; .drawio writes by any process other than draw.io are caught by a file-write audit log and score 0.
Proposed partial credit
45% edges; 30% alignment; 15% equal spacing; 10% others untouched
R11.03.01.01-T08 Fix BPMN notation errors in a Visio process diagram
Original instruction
output/claims_bpmn.vsdx (Visio BPMN template) fails Check Diagram with 9 issues: sequence flows across pools, a split gateway without a join, a task with no outgoing flow, a missing end event, and mislabeled events. Fix them so Visio's BPMN rule validation reports 0 issues without removing activities.
- Platforms
win· Windows- Applications
- Visio
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropclick· Point & clickvis· Visual judgment- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro apphidden· Hidden/stale state- Proposed difficulty
- 4 / 5
Proposed verifier
Re-run Visio validation via verifier automation: 0 issues. Parse VSDX: cross-pool links are Message Flows; all 14 activities still present; one join gateway added; end events present on each terminal path. Visio VBA and COM automation are disabled by policy and the agent has no terminal or Python, so the .vsdx must be saved by Visio (docProps/app.xml checked).
Proposed partial credit
50% issues resolved; 25% message flows correct; 25% activities preserved
R11.03.01.01-T09 Import a Visio file into Lucidchart and repair connectors
Original instruction
Import legacy/onboarding.vsdx into Lucidchart. After import, 6 connectors are detached and 2 shapes lost their text. Reattach connectors to the correct shapes (per the original PNG legacy/onboarding.png), restore the text, and export output/onboarding_fixed.pdf and .vsdx.
- Platforms
web· Any OS (web)- Applications
- Lucidchart
- Interfaces
brow· Browser- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropvis· Visual judgmenttype· Text & forms- Work product
WP03· Visual layout- Conditions
multiapp· Multi-app- Proposed difficulty
- 3 / 5
Proposed verifier
Exported VSDX: graph isomorphic to original incl. 6 repaired edges glued; 2 restored labels exact; PDF exported. No Lucid API key or OAuth app is issued to the agent, so edits must come from Lucidchart browser sessions (checked in document revision history).
Proposed partial credit
50% connectors repaired; 25% text restored; 25% exports
R11.03.01.01-T10 Turn an eligibility policy into a decision-tree flowchart
Original instruction
From policy_eligibility.pdf (8 rules with thresholds on age, income, residency), draw a draw.io decision flowchart that routes any applicant to Eligible, Refer or Ineligible, with each diamond testing one condition and Yes/No labeled branches. Save output/eligibility.drawio.
- Platforms
mac· macOSwin· Windowslin· Linuxweb· Any OS (web)- Applications
- draw.io
- Interfaces
desk· Desktop GUIbrow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & droptype· Text & formsgeom· Precise geometry- Work product
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Parse graph; verifier interprets diamonds (condition text parsed with a small grammar) and walks 200 synthetic applicants: outcome agrees with policy reference for ≥98%. All diamonds have exactly two labeled outgoing edges; no orphan nodes. Agent has no terminal or text editor; .drawio writes by any process other than draw.io are caught by a file-write audit log and score 0.
Proposed partial credit
20% structure valid; 60% routing accuracy; 20% labels
R11.03.01.01-T11 Generate a multi-page org chart with photos from Excel
Original instruction
Use Visio's Organization Chart wizard with staff.xlsx (name, title, reports_to, dept, photo path) to build a 3-level chart with photos, then split it so each department is on its own page with off-page references back to the executive page. Save output/org_pages.vsdx.
- Platforms
win· Windows- Applications
- Visio · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickdrag· Drag & droptype· Text & forms- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro appmultiapp· Multi-app- Proposed difficulty
- 4 / 5
Proposed verifier
Parse VSDX: pages = 1 executive + 5 departments; each department page holds exactly its staff; images embedded for all with photos; off-page reference shapes paired across pages; hierarchy edges match Excel. Visio VBA and COM automation are disabled by policy and the agent has no terminal or Python, so the .vsdx must be saved by Visio (docProps/app.xml checked).
Proposed partial credit
25% wizard import; 30% department pages; 20% photos; 25% off-page refs and hierarchy
R11.03.01.01-T12 Restyle a flowchart to the brand diagram standard
Original instruction
In OmniGraffle, restyle output/support_flow.graffle per diagram_style.pdf: process boxes 160x60 rounded 8 px fill #F2F5FA stroke #1F3A5F 1.5 pt, decisions in #FFF4D6, Inter 12 pt, orthogonal connectors with line hops, 24 px grid alignment. Keep topology unchanged.
- Platforms
mac· macOS- Applications
- OmniGraffle
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickgeom· Precise geometryvis· Visual judgment- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Parse .graffle: per shape size/corner/fill/stroke/font match spec (tolerance ±1 px, ΔE≤2); line type orthogonal with hops; positions on 24 px grid; graph unchanged vs input. Omni Automation and AppleScript access to OmniGraffle are disabled (TCC denies Automation) and the agent has no terminal; the file must be saved by OmniGraffle.
Proposed partial credit
40% shape styles; 20% fonts; 20% connectors; 20% grid and topology kept
R11.03.01.01-T13 Validate and simulate a process in Bizagi Modeler
Original instruction
In Bizagi Modeler open models/claims.bpm. Fix validation errors, then set up Process Validation and Time Analysis simulation using sim_params.xlsx (arrival every 10 min, task durations, gateway probabilities), run 30 days and record average cycle time and resource utilization of Adjuster in sim_report.txt.
- Platforms
win· Windows- Applications
- Bizagi Modeler
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
click· Point & clicktype· Text & formsnav· Pan/zoom/scroll- Work product
WP07· Simulation result- Conditions
multiapp· Multi-apptutorial· Follow docs/tutorial- Proposed difficulty
- 5 / 5
Proposed verifier
Exported BPMN with simulation extension elements equal to the param sheet; no validation errors; sim_report.txt values within 5% of gold simulation (seeded) for cycle time and Adjuster utilization. Agent has no terminal or archive tools; .bpm writes by any process other than Bizagi Modeler are caught by a file-write audit log and score 0.
Proposed partial credit
20% validation fixed; 35% parameters entered; 30% results within tolerance; 15% report
R11.03.01.01-T14 Redraw a whiteboard photo as a clean flowchart
Original instruction
input/whiteboard.jpg is a phone photo of a hand-drawn incident-response flow (11 nodes, some arrows crossing, 2 decisions). Recreate it in draw.io as a clean top-to-bottom flowchart with correct shapes and labels and no crossing edges. Save output/ir_flow.drawio.
- Platforms
mac· macOSwin· Windowslin· Linuxweb· Any OS (web)- Applications
- draw.io
- Interfaces
desk· Desktop GUIbrow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
vis· Visual judgmentdrag· Drag & droptype· Text & forms- Work product
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Graph vs gold transcription: node labels (normalized fuzzy match ≥0.9), edges isomorphic; decision shapes rhombus; layout: every edge points downward or sideways, 0 edge crossings. Agent has no terminal or text editor; .drawio writes by any process other than draw.io are caught by a file-write audit log and score 0.
Proposed partial credit
30% nodes and labels; 35% edges; 15% shape types; 20% clean layout
R11.03.01.01-T15 Link an org chart to a sheet and highlight open roles
Original instruction
In Lucidchart, link the org chart in 'Sales Org' to Google Sheet sales_roster (data linking) and add conditional formatting: shapes with status 'Open' get a dashed red border and the text 'OPEN ROLE'. Refresh after the sheet is updated with 2 new hires.
- Platforms
web· Any OS (web)- Applications
- Lucidchart · Google Sheets
- Interfaces
brow· Browser- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickmulti· Cross-app- Work product
WP03· Visual layout- Conditions
multiapp· Multi-app- Proposed difficulty
- 3 / 5
Proposed verifier
Lucid export: shapes linked to sheet rows; conditional rule present; after refresh 2 previously open roles show names and no red border; remaining open roles (4) styled per rule. No Lucid API key or Google Sheets API token is issued to the agent, so linking and refresh must come from browser sessions.
Proposed partial credit
35% data link; 30% conditional formatting; 35% refreshed state correct
R11.03.01.01-T16 Untangle a messy 40-node flowchart without changing its logic
Original instruction
output/fulfilment.vsdx is a 40-node flowchart with crossing connectors and uneven spacing. Re-layout it top-to-bottom so there are no connector crossings, shapes snap to grid with uniform 30 mm vertical spacing between ranks, and decision branches go right for 'No'. Keep the graph identical.
- Platforms
win· Windows- Applications
- Visio
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & dropgeom· Precise geometryvis· Visual judgment- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Parse VSDX geometry: connector polylines have 0 intersections; rank spacing 30±1 mm; 'No' edges exit to the right; graph and labels identical to input. Visio VBA and COM automation are disabled by policy and the agent has no terminal or Python, so the .vsdx must be saved by Visio (docProps/app.xml checked).
Proposed partial credit
40% zero crossings; 20% spacing; 20% No-branch convention; 20% graph unchanged
R11.03.01.01-T17 Generate the flowchart with Graphviz (control pair of T02)
Original instruction
Control pair of T02. Without a diagram GUI, write flow.dot describing the same flowchart (shapes, labels, Yes/No branches) and render output/flow_cli.svg and output/flow_cli.png with the dot CLI.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Graphviz
- 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
Parse flow.dot: graph isomorphic to reference with labels; decision node shape diamond; SVG/PNG rendered.
Proposed partial credit
60% graph; 20% shapes; 20% renders
R11.03.01.01-T18 Model an order-to-cash collaboration with compensation and run it
Original instruction
In Camunda Modeler model 'Order to Cash' as a collaboration with pools Customer (collapsed), Sales and Warehouse: message flows for order and shipment, an event subprocess for cancellation that triggers compensation of 'Reserve stock', and data objects Invoice. Deploy and run token simulation for both normal and cancel paths.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Camunda Modeler
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
drag· Drag & dropclick· Point & clicktype· Text & forms- Work product
WP06· Code/engineering- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
BPMN XML: participants, message flows only between pools, event subprocess with message start, compensation boundary + handler linked by association, data objects; bpmnlint clean. Engine: both scenario instances complete with expected activity history. Agent has no terminal, text editor or HTTP client, so the BPMN must be saved by Camunda Modeler (exporter attribute checked) and deployed from its Deploy dialog.
Proposed partial credit
25% pools and message flows; 30% event subprocess and compensation; 15% lint; 30% both paths run
R11.03.01.01-T19 Build a collapsible org chart with containers in draw.io
Original instruction
In draw.io create output/research_org.drawio for the 30 people in research_staff.csv: one container per lab (5 labs) holding its members, tree layout inside each lab under its lead, and the director above all labs. Containers must be collapsible and labeled with lab name and headcount.
- Platforms
mac· macOSwin· Windowslin· Linuxweb· Any OS (web)- Applications
- draw.io
- Interfaces
desk· Desktop GUIbrow· Browser- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
drag· Drag & droptype· Text & formsgeom· Precise geometry- Work product
WP03· Visual layout- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Parse XML: 5 container cells (collapsible) whose children equal lab members; labels include correct headcount; reporting edges match CSV; leads above members (y order). Agent has no terminal or text editor; .drawio writes by any process other than draw.io are caught by a file-write audit log and score 0.
Proposed partial credit
30% containers and membership; 35% reporting edges; 20% layout order; 15% labels
R11.03.01.01-T20 Overlay KPI data graphics on a Visio process map
Original instruction
In Visio link the 9 process steps of output/ap_process.vsdx to ap_kpis.xlsx (step id, cycle time h, error rate) with Link Data to Shapes, then apply a data graphic: cycle time as text callout and error rate as colored icon (green <2%, amber 2–5%, red >5%). Save.
- Platforms
win· Windows- Applications
- Visio · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickdrag· Drag & drop- Work product
WP03· Visual layout- Conditions
licensed· Licensed pro appmultiapp· Multi-app- Proposed difficulty
- 3 / 5
Proposed verifier
Parse VSDX: DataRecordSet from ap_kpis.xlsx; each step shape linked to correct row (Prop.ID); data graphic master applied; icon set thresholds as specified; rendered PNG export shows correct color per step. Visio VBA and COM automation are disabled by policy and the agent has no terminal or Python, so the .vsdx must be saved by Visio (docProps/app.xml checked).
Proposed partial credit
40% correct data links; 30% data graphic applied; 30% thresholds/colors
Mapped research references
No benchmark reference is assigned to this class in the current mapping.