Mouse Gestures That Answer Back

Mouse gestures are at their best when they feel physical. You hold a button, move through space, release, and something happens. The trap is that the software can make that feel either magical or weirdly silent.

This release adds a small bit of magic: user-level mouse shortcut rules can opt into a matrix completion animation. When a shape gesture matches, Lattices takes the captured gesture path, cleans it up, replays it inside a tiny 3x3 matrix, then snaps into a confirmation glyph and label.

It is deliberately small. It is not a mascot, not a system-wide animation layer, and not part of whether the action succeeds. It is feedback: the app showing that it understood your motion.

Your Gesture, Cleaner

The first version of shape feedback had a tempting simplification: recognize a shape, then replay a perfect canonical version of it.

That was too clean.

The better version keeps the personality of the actual gesture. Lattices already captures the path points for recognition and drawing. The matrix renderer uses those points as source material:

1. simplify the captured path
2. smooth out tiny hand wobble
3. scale the result into a compact matrix
4. replay the path quickly
5. pulse the active cells
6. show the action label

So the animation is not a literal raw trail, but it is still yours. The gesture becomes UI.

Why A Matrix

The Lattices mark is a 3x3 grid of rounded cells: an L made from the left column and bottom row. That is already the product's visual language for spatial control.

Putting gesture confirmation into the same grammar makes it feel native to the app. A back-button down-then-left gesture that sends Return can resolve as a small matrix movement, a corner pulse, and a Return tag. The shape does not need a character or a long flourish. It just needs to feel recognized.

User-Level, Opt-In

This is intentionally not a committed default for every gesture.

Mouse shortcuts live in:

~/.lattices/mouse-shortcuts.json

That file is machine-local preference data. The app code owns the schema, recognizer, and fallback defaults, but personal shortcut mappings belong to the user.

The matrix completion renderer follows the same rule. A shortcut opts in with a visual block:

{
  "id": "back-down-left-enter",
  "enabled": true,
  "device": "any",
  "trigger": {
    "button": "back",
    "kind": "shape",
    "shape": "l-shape-down-left"
  },
  "action": {
    "type": "shortcut.send",
    "shortcut": {
      "key": "enter",
      "keyCode": 36,
      "modifiers": []
    }
  },
  "visual": {
    "renderer": "matrix",
    "theme": "lattices",
    "message": "Return"
  }
}

No visual block, no embellishment. The action still runs.

The Boundary

The important engineering boundary is that visuals never decide behavior.

Gesture recognition, rule matching, and action dispatch remain the fast path. Matrix completion is downstream of that. If the renderer changes, misses a frame, or is disabled, the matched action is unaffected.

That keeps the feature in the right emotional category: delightful feedback, not fragile input plumbing.

What Shipped

This release includes:

• sub-threshold middle-button movement passthrough, so Chrome middle-click tab close keeps working while Lattices watches for real gestures
• clearer documentation that mouse shortcuts are user-level config, not project defaults
• an opt-in native matrix visual renderer for shape gesture completions
• a local Return gesture example: back button, down then left, release

The small interaction is the point. When a gesture completes, Lattices can now answer back in its own visual language.