Interaction models

North Star: predictable and forgiving: the same gesture does the same thing everywhere, and nothing is silent or irreversible (SUB-21, SUB-28).

Worked patterns applied the same way across hosts, tools and agents (DEV-08). Each model is one short page: the requirements it cites, a Mermaid sequence diagram, and a pointer to its reference implementation. Where an interaction is already a requirement, the model cites it rather than restating it.

Model	Written in	Status
Selection and tool offering	Phase 1	Requirements gathered
Tool invocation and forms	Phase 1	Requirements gathered
Unavailable capabilities	Phase 1	Requirements gathered
Undo	Phase 2	Written
Conflicts	Phase 2	Written
Long-running operations	Phase 2	Requirements gathered
Check overrides	Phase 2	Requirements gathered
Time controller	Phase 3	Written
Annotations	Phase 3	Written (forms; drawing and editing on the map)
Track operations	Phase 3	Written (part A: move, rotate, split, merge; part B1: shear, stretch; part B2: jumps, smoothing, marked fixes; part B3: single fixes, dead reckoning)
Labels, symbols and arrows	Phase 3	Written (intervals per track, aligned to the clock as legacy Debrief; thinning; a fix's own label or symbol)
Tote	Phase 3	Written (part A: primary and secondaries, the core measures, latency)
Timeline and narrative	Phase 3	Written (part A: narrative entries, REP and .dpf import, the Timeline following current time, filter, search, editing, scale)
Collation	Phase 3	Written (coverage, platforms across items, source preference per period, new data, taking it, review flags)
Exported images	Phase 3	Written (the map as an image with an ID naming the plot and the revision, the footer, the export log; finding an image's source, read-only)
Keyboard equivalents	Phase 3	Partly written (annotations, tracks, the Tote, the Timeline, collation, exported images)

Analysts review the Phase 1 models before they are built.