@maccesar/titools 4.8.0 → 4.10.0

package/README.md

@@ -177,7 +177,7 @@ The knowledge index is based on the latest Titanium SDK documentation. If your p
 | ti-expert    | Architecture and implementation     | Starting point for most tasks           |
 | purgetss     | Utility-first styling               | UI styling and animations               |
 | ti-ui        | UI/UX patterns                      | Complex layouts, ListViews, platform UI |
-| ti-game      | 2D games with the `ti.game` module  | Sprites, physics, collisions, text HUDs, particles, camera |
+| ti-game      | 2D games with the `ti.game` module  | Sprites, physics, collisions, raycasts, pathfinding, text HUDs, particles, camera |
 | ti-api       | Complete Titanium API reference     | Looking up properties, methods, events  |
 | ti-guides    | SDK fundamentals                    | tiapp.xml, Hyperloop, distribution, JDK/Xcode compatibility, release notes |
 | ti-howtos    | Native feature integration          | Location, push, media, platform APIs    |
@@ -436,8 +436,11 @@ When it activates:
 - Sprites, sprite sheets, TexturePacker atlases, frame animations, native tweens
 - Physics: gravity and velocity, solid platforms, one-way floors, bouncing, the arcade car model, Newtonian thrust
 - Collision groups, invisible trigger zones, enter/exit events, swept AABB for fast bullets, hitbox tuning
+- Raycasts for line of sight, ledge probes and hitscan shots
+- A* pathfinding (`findPath`) around tagged obstacles, feeding native patrol paths (`followPath`) and animation chaining (`play(name, { then })`)
 - Bitmap-font text sprites and screen-fixed HUDs inside the scene
-- Particle emitters, Verlet ropes, camera follow/zoom/shake/effects, low-latency sound
+- Particle emitters, Verlet ropes, camera follow/zoom/shake/effects, `scrollFactor` parallax, low-latency sound
+- Game-clock timers that obey pause and slow motion
 - Drag & drop, pinch, rotate, multitouch controls
 
 Example prompts:
@@ -448,6 +451,8 @@ Example prompts:
 "How do I make a top-down racer that drifts?"
 "Why does my game stutter when I move sprites from setInterval?"
 "Add a score HUD that stays put while the camera follows the player."
+"Make the guard walk a patrol route and turn around before the ledge."
+"Tap to walk, but the player should go around the tree instead of through it."
 ```
 
 Key rules:
@@ -457,6 +462,7 @@ Key rules:
 - A sprite with size but no sheet is an invisible trigger (scores, goals, walls)
 - Add a whole level with one `gameView.add([...])` call
 - HUDs are `Game.createText` sprites with `screenFixed: true`, not `Ti.UI.Label` overlays
+- Timers that drive game logic belong on the game clock (`gameView.every`), so a paused game stops spawning
 
 ---
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@maccesar/titools",
-  "version": "4.8.0",
+  "version": "4.10.0",
   "description": "Titanium SDK skills and agents for AI coding assistants (Claude Code, Gemini CLI, Codex CLI)",
   "main": "lib/index.js",
   "type": "module",

package/skills/ti-game/SKILL.md

@@ -1,13 +1,13 @@
 ---
 name: ti-game
-description: "Use when building or reviewing 2D games with the ti.game module for Titanium SDK (Android and iOS) — sprites, sprite sheets, frame animations, tweens, physics (gravity, solid platforms, carMode drifting, thrust), collision groups, trigger zones, swept AABB, bitmap-font text and HUDs, particles, Verlet ropes, camera follow/zoom/shake, low-latency sound, drag & drop, multitouch. AUTO-DETECT: if tiapp.xml declares ti.game or any file calls require('ti.game'), invoke BEFORE writing ANY game code. Also trigger on: game, juego, sprite, spritesheet, platformer, endless runner, top-down, shooter, racer, flappy, breakout, rhythm game, card game, puzzle, collision, hitbox, parallax, particles, HUD, score, game loop, 60 fps, GameView. ti.game is NOT Phaser, Godot or Box2D: you never write a game loop, never move a sprite from setInterval, and several familiar features (tilemap layer, sprite parenting, raycast, joystick) do not exist yet — check references/roadmap.md before inventing an API."
+description: "Use when building or reviewing 2D games with the ti.game module for Titanium SDK (Android and iOS) — sprites, sprite sheets, animations, tweens, physics (gravity, solids, carMode drifting, thrust), collision groups, trigger zones, swept AABB, raycast, paths, bitmap text, particles, ropes, camera follow/zoom/shake, parallax, sound, drag & drop, multitouch. AUTO-DETECT: if tiapp.xml declares ti.game or any file calls require('ti.game'), invoke BEFORE writing ANY game code. Also trigger on: game, juego, sprite, spritesheet, platformer, endless runner, top-down, shooter, racer, flappy, breakout, rhythm, cards, puzzle, collision, hitbox, raycast, line of sight, pathfinding, maze, patrol, parallax, particles, HUD, score, game loop, 60 fps, GameView. ti.game is NOT Phaser, Godot or Box2D: you never write a game loop, never move a sprite from setInterval, and several familiar features (tilemap layer, sprite parenting, joystick, word wrap) do not exist yet — check references/roadmap.md before inventing an API."
 argument-hint: "[genre or feature]"
 allowed-tools: Read, Grep, Glob, Edit, Write, Bash(node *)
 ---
 
 # ti.game — 2D game engine for Titanium SDK
 
-`ti.game` is a native OpenGL ES 2.0 sprite engine exposed to JavaScript, written and maintained by **Michael Gangolf (m1ga)** — upstream is https://github.com/m1ga/ti.game. This skill tracks upstream `main` as of **2026-08-18** — the manifest still reads `0.3.0`, but the text engine, `screenFixed`, `swept` collisions and `collisionend` landed after it, so a version number alone will not tell you what a checkout has. When upstream and any fork disagree, upstream wins. Same JS API on Android and iOS — the iOS side is a class-per-class Obj-C twin of the Android engine.
+`ti.game` is a native OpenGL ES 2.0 sprite engine exposed to JavaScript, written and maintained by **Michael Gangolf (m1ga)** — upstream is https://github.com/m1ga/ti.game. This skill tracks upstream `main` as of **2026-08-20**, manifest **`0.4.0`** — that bump is the first in a long run of feature landings (the text engine, `screenFixed`, `swept` collisions, `collisionend`, `followPath`, animation chaining, `raycast`, `findPath`, game-clock timers, `scrollFactor` parallax and the `multiply`/`screen` blend modes all shipped while it still said `0.3.0`), so anything still calling itself 0.3.0 has none of them. When upstream and any fork disagree, upstream wins. Same JS API on Android and iOS — the iOS side is a class-per-class Obj-C twin of the Android engine.
 
 Works in **both Classic and Alloy** projects. Nothing here depends on PurgeTSS.
 
@@ -29,11 +29,17 @@ The practical consequence, and the single biggest mistake to avoid:
 | A car that drives and drifts | your own physics | `carMode: true` + `throttle`/`steering` |
 | A ship with inertia | vector math per frame | `thrust` + `angularVelocity` |
 | Drag a sprite with the finger | `touchmove` handler | `draggable: true` |
-| Scrolling background | timer repositioning copies | `wrapX` / `wrapShift` + `velocityX` |
+| Scrolling background under a fixed camera | timer repositioning copies | `wrapX` / `wrapShift` + `velocityX` |
+| Parallax layers under a moving camera | hand-scrolled layers | `scrollFactor` on each layer |
 | Camera that follows the player | timer writing `cameraX` | `gameView.follow(sprite, options)` |
 | A score, a title, a label | `Ti.UI.Label` overlays | `Game.createText({ text, screenFixed: true })` |
 | A bullet fast enough to skip past a thin wall | shrinking the time step | `swept: true` on the bullet |
 | To know the player *left* a zone | polling on a timer | the `collisionend` event |
+| An enemy on a fixed patrol route | chained `animate()` legs | `sprite.followPath(points, { loop })` |
+| Play an attack, then go back to idle | an `animationcomplete` handler | `play('attack', { then: 'idle' })` |
+| Line of sight, a ledge probe, a hitscan shot | an invisible probe sprite | `gameView.raycast(x0, y0, x1, y1, ['group'])` |
+| Walking around obstacles to a tapped point | your own A* in JS, or a straight line through the wall | `gameView.findPath(from, to, opts)` → `sprite.followPath(path)` |
+| A spawn wave or an AI tick that respects pause | `setInterval` | `gameView.every(ms, cb)` |
 | Smoke, sparks, explosions | pooled sprites in JS | `Game.createEmitter(...)` |
 | A swinging chain or cape | verlet solver in JS | `Game.createRope(...)` |
 | Slow motion or a pause that still draws | pausing your own timers | `gameView.timeScale = 0.5` / `0` |
@@ -128,6 +134,10 @@ The enter/exit pair is the trigger lifecycle other engines call `OnTriggerEnter`
 
 A fast mover can slip through a thin target between frames: it never overlaps on any single frame, so the discrete test misses. `swept: true` on the *mover* tests its movement as a path — `collision` fires for whatever the path crossed, and `solidWith` walls stop it at the impact point instead of letting it teleport through.
 
+When the question is not "did we touch?" but "what is over there?", the answer is `gameView.raycast(x0, y0, x1, y1, ['group'])` — line of sight, a ledge probe ahead of a walker's feet, a hitscan shot. It returns `null` or the nearest hit (`{ x, y, distance, group, sprite, normal }`), needs nothing but a `collisionGroup` on the targets, and replaces the old trick of parking an invisible probe sprite where the ray would land. It is a discrete query: call it from a decision timer or a tap handler, never every frame.
+
+And when the question is "how do I get over there?", `gameView.findPath(from, to, { cellSize, groups, clearance, bounds })` runs grid A* around the same tagged sprites and hands back waypoints for `followPath` — tap-to-walk that rounds a tree, a hound re-routing to the player on an 800 ms timer. Also discrete: the walk itself runs natively between queries, so pathfind on taps and AI ticks, never per frame. Pass `clearance` of about half the walker's width, since the route is a line for the sprite's **center**.
+
 **A sprite with `width`/`height` but no `sheet` renders nothing and works as an invisible trigger.** Score zones, goals, checkpoints, ceilings, kill floors and walls are all built this way in the official demos. This is the idiomatic way to add logic to a scene without art.
 
 Tune fairness with `hitboxScale` (art rarely fills its frame; slightly small hitboxes feel better) and `hitboxShape: 'circle'` for balls and rocks — circles also resolve against solids along the contact normal, so they bounce off corners diagonally instead of like a box. Turn on `debug: true` per sprite, or on the GameView for everything, to see green collision AABBs, blue touch bounds and the orange anchor dot.
@@ -135,7 +145,7 @@ Tune fairness with `hitboxScale` (art rarely fills its frame; slightly small hit
 ## Performance rules that actually matter
 
 - **One texture = one draw call.** The batcher accumulates up to 1000 quads and flushes on texture switch. A scene whose sprites all share one sheet renders in a single draw call regardless of sprite count. Pack art into as few sheets as possible; reuse one white particle texture and tint it at runtime.
-- **Know what else breaks the batch**, since none of it is visible on screen: every `'normal'` ↔ `'add'` transition in draw order flushes (group additive sprites into their own `zIndex` band so it happens twice a frame, not constantly); each glowing sprite costs 2 extra draw calls per frame, plus 2 more while a `flash()` runs; ropes, skid marks and debug overlays always draw with normal blending, so interleaving them with additive content flushes too.
+- **Know what else breaks the batch**, since none of it is visible on screen: every blend-mode transition in draw order flushes — there are four (`'normal'`, `'add'`, `'multiply'`, `'screen'`), so group same-blend sprites into their own `zIndex` band and the frame switches a handful of times instead of constantly; each glowing sprite costs 2 extra draw calls per frame, plus 2 more while a `flash()` runs; ropes, skid marks and debug overlays always draw with normal blending, so interleaving them with non-normal content flushes too.
 - **Bulk-add a level.** `gameView.add([spriteA, spriteB, emitter, rope])` crosses the bridge once and commits under one scene lock. Collect a level's objects in an array and add them in one call; `add(object)` still works for single objects.
 - **Keep collision group lists targeted.** Cost is O(colliders × candidates) per frame. A bullet checking `['asteroid']` tests 5 sprites, not the whole scene.
 - **Pool, don't create.** Bullets, notes, obstacles and enemies are created once (usually 3–10 of them), then recycled by setting `visible = false` and zeroing velocity. Creating sprites mid-game costs a bridge hop and a scene lock.
@@ -161,7 +171,9 @@ Standard Titanium touch events do not reach the GameView, but ordinary Titanium
 
 ## Clean up what you created
 
-The render loop follows the activity lifecycle automatically, but **your timers and sounds do not**. Every demo that starts an interval or plays looping music stops it on window close. In Alloy, wire this into the controller's `cleanup()`.
+The render loop follows the activity lifecycle automatically, but **your JS timers and sounds do not**. Every demo that starts an interval or plays looping music stops it on window close. In Alloy, wire this into the controller's `cleanup()`.
+
+Game-clock timers (`gameView.after` / `every`) are the exception that proves the rule: they run on the engine's clock, so they freeze at `timeScale: 0` and pause with the render loop instead of firing into a dead scene. That is why spawn waves and AI ticks belong there. A `setInterval` keeps running through a pause, which is right for a real-time countdown and wrong for everything else.
 
 ```javascript
 win.addEventListener('close', () => {
@@ -180,7 +192,7 @@ Read the one that matches the task instead of guessing — the API surface below
 | Need | Reference |
 | --- | --- |
 | Exact properties, methods, events, defaults for every object | [api.md](references/api.md) |
-| Working code per genre: flappy, platformer, top-down, racer, asteroids, endless runner, rhythm, cards, point & click, particles, ropes, camera | [recipes.md](references/recipes.md) |
+| Working code per genre: flappy, platformer, top-down, racer, asteroids, endless runner, rhythm, cards, point & click, particles, ropes, camera — plus patrol paths, raycast probes, pathfinding, parallax and game-clock timers | [recipes.md](references/recipes.md) |
 | Install the module, tiapp.xml, Alloy vs Classic wiring, assets, sprite sheets, atlases, lifecycle, testing | [project-setup.md](references/project-setup.md) |
 | What does NOT exist yet, what is coming, and how to work around it today | [roadmap.md](references/roadmap.md) |
 
@@ -195,4 +207,5 @@ Read the one that matches the task instead of guessing — the API surface below
 7. Is every `createSound`, `setInterval` and `setTimeout` stopped when the window closes?
 8. Is the HUD `Game.createText` with `screenFixed: true`, rather than a stack of overlaid labels?
 9. Does anything travel far enough per frame to tunnel? Give it `swept: true`.
-10. Are you using an API that actually exists? Anything involving tilemaps, sprite parenting, raycasting, joysticks or `blend: 'multiply'` — check [roadmap.md](references/roadmap.md) first.
+10. Does a timer drive game logic (spawn waves, AI ticks, respawns)? Move it to `gameView.every`/`after` so it obeys pause and slow motion.
+11. Are you using an API that actually exists? Anything involving tilemaps, sprite parenting, joysticks, gamepads or word-wrapped text — check [roadmap.md](references/roadmap.md) first.

package/skills/ti-game/references/api.md

@@ -1,6 +1,6 @@
 # ti.game API reference
 
-Complete JS surface of `ti.game` as of upstream `main` on 2026-08-18, verified against the module source (`android/src/ti/game/*.java` and `engine/*.java`, mirrored by `ios/Classes/TG*`). Defaults come from the engine fields, not from prose. The manifest still reads `0.3.0`; the text engine (`createFont`/`createText`), `screenFixed`, `swept` and `collisionend` landed after it and are not in any numbered release yet.
+Complete JS surface of `ti.game` as of upstream `main` on 2026-08-20, verified against the module source (`android/src/ti/game/*.java` and `engine/*.java`, mirrored by `ios/Classes/TG*`). Defaults come from the engine fields, not from prose. The manifest reads **`0.4.0`** since 2026-08-20, and that bump matters: the text engine (`createFont`/`createText`), `screenFixed`, `swept`, `collisionend`, `followPath`, animation chaining, `raycast`, `findPath`, the game-clock timers, `scrollFactor` and the `multiply`/`screen` blend modes all landed while the manifest still said `0.3.0`, so a build calling itself 0.3.0 has none of them.
 
 Every property listed is **live**: reading returns the current native value, even mid-drag or mid-tween. Every property can also be passed to its `create*` factory. All durations crossing the JS boundary are **milliseconds** (the engine converts to seconds internally).
 
@@ -57,6 +57,11 @@ A normal Titanium view — add it to a window or any container, size it with the
 | `follow(sprite, options)` | method | — | Native dead-zone follow, see below |
 | `stopFollow()` | method | — | Stop following; the camera stays where it is |
 | `shake({ strength, duration })` | method | `12` px, `400` ms | Detuned-sine rumble on the projection only — follow, bounds and touch mapping are unaffected |
+| `raycast(x0, y0, x1, y1, groups)` | method | — | One-shot nearest-hit query along the segment against **visible, non-`screenFixed`** sprites carrying a `collisionGroup` in `groups` (omit for any tagged sprite). Returns `null` for a clear ray, else `{ x, y, distance, group, sprite, normal: { x, y } }`. Rect hitboxes are tested as their AABB, circle hitboxes exactly; a ray starting inside a hitbox reports it at distance 0 |
+| `findPath(from, to, options)` | method | — | Grid A\* from `from` to `to` (`{ x, y }` world points) around the same sprites `raycast` sees — **visible, non-`screenFixed`**, carrying a `collisionGroup` in `options.groups` (omit for any tagged sprite). Returns an array of `{ x, y }` waypoints ready for `sprite.followPath()`, or `null` when no route exists. Options below |
+| `after(ms, callback)` | method | — | Runs `callback` once after `ms` of **game time** — the delay stretches under `timeScale` slow motion, freezes at `0` and pauses with the render loop, unlike `setTimeout`. Returns an int id. The callback receives `{ id }` |
+| `every(ms, callback)` | method | — | Like `after()`, repeating until cancelled. Fires at most once per frame, and restarts its interval after a stall instead of bursting to catch up |
+| `cancelTimer(id)` | method | — | Cancels a timer from `after()` / `every()` |
 | `cameraEffect` | string | `'none'` | Fullscreen shader pass: `'none'`, `'tint'`, `'glitch'`. With `'none'` the extra pass is skipped entirely |
 | `cameraTint` | string | — | Color for the `'tint'` effect, e.g. `'#4f8'` (night vision, poison, flashback) |
 | `cameraEffectIntensity` | float | `1` | 0..1 — tint mix or glitch amount |
@@ -64,6 +69,8 @@ A normal Titanium view — add it to a window or any container, size it with the
 
 Events: `press`, `tap`, `release` — fired for **every** touch anywhere on the view, with payload `x`, `y` in scene coordinates (tap-anywhere controls, flappy-style). And `resize` with payload `width`, `height` — the real surface size in pixels.
 
+Plus `timer` with payload `id`, fired only for `after()` / `every()` calls made **without** a callback — pass a callback or listen for the event, not both.
+
 ### `follow(sprite, options)`
 
 | Option | Default | Effect |
@@ -79,6 +86,31 @@ Vertical follow is always on. **Horizontal follow is off until you pass `leftMar
 
 Each call **resets every option to its default** before applying what you passed. Calling `follow(sprite)` with no options after configuring it wipes the configuration.
 
+### `findPath(from, to, options)`
+
+| Option | Default | Effect |
+| --- | --- | --- |
+| `cellSize` | `32` | Grid resolution in px. Match it to the tile size on a tile map so the grid lines up with the walls; a coarser grid is faster and blockier |
+| `groups` | all tagged sprites | Array of `collisionGroup` names to treat as obstacles |
+| `clearance` | `0` | Extra obstacle inflation in px. About half the walker's width keeps it from scraping corners, since the path is a line for the sprite's **center** |
+| `bounds` | the whole surface | `{ minX, minY, maxX, maxY }` search rect — also the walkable area, so a floor clamp goes here |
+| `diagonals` | `true` | Octile moves. Diagonals never cut a blocked corner |
+| `simplify` | `true` | Line-of-sight string pulling: the cell chain collapses to the few corner waypoints `followPath` needs. `false` returns every grid cell (useful to visualize or debug the route) |
+
+Waypoints come back with the **exact** start and goal as first and last point, so the walk begins under the sprite's feet and ends on the tapped pixel. A blocked start or goal snaps outward to the nearest free cell (up to 4 rings), which is why tapping an obstacle walks to its edge instead of returning `null`. `null` means: no route, degenerate `bounds`, `cellSize <= 0`, or a grid over ~1M cells (a runaway `bounds`/`cellSize` combination fails the query rather than allocating).
+
+Everything is built per call on the calling thread — no persistent grid, no GL work. Like `raycast` it is a **discrete query**: run it on taps and AI timers (`gameView.every`), never per frame.
+
+```javascript
+const path = gameView.findPath(
+	{ x: player.x, y: player.y },
+	{ x: e.x, y: e.y },
+	{ cellSize: TILE, groups: ['wall'], clearance: player.width * 0.35, bounds: WALK_BOUNDS });
+if (path) {
+	player.followPath(path, { speed: 160 });
+}
+```
+
 ## SpriteSheet
 
 Create one sheet per texture and share it across every sprite that uses it — sprites sharing a sheet batch into a single draw call.
@@ -88,7 +120,7 @@ Create one sheet per texture and share it across every sprite that uses it — s
 | `image` | — | Path to the texture, relative to the resources root (`'assets/hero.png'`) |
 | `frameWidth` / `frameHeight` | — | Grid mode: cell size in source pixels. Frames are numbered left-to-right, top-to-bottom, starting at 0 |
 | `atlas` | — | TexturePacker JSON (hash or array format) instead of a grid |
-| `smoothing` | `true` | `false` = GL_NEAREST, the pixel-art setting |
+| `smoothing` | `true` | `false` = GL_NEAREST, the pixel-art setting. On a **grid** sheet, `true` also insets the interior frame UVs by half a texel so magnified edges cannot sample the neighbouring frame; outer edges keep the exact 0..1 range so `tileRepeat` still wraps. Atlas frames come from the JSON untouched |
 | `repeat` | `false` | GL_REPEAT wrap, required by sprites using `tileRepeat`. Needs power-of-two texture dimensions |
 
 | Member | Notes |
@@ -117,6 +149,7 @@ Textures upload to the GPU lazily on first use, from the render thread, and are
 | `ySort` | `false` | Within the same `zIndex`, sort by the sprite's **bottom edge** — top-down depth (walk behind a tree, in front of it below) |
 | `flipX`, `flipY` | `false` | Mirror the drawn frame only. Position, anchor, physics and hit testing are untouched |
 | `screenFixed` | `false` | `x`/`y` become **surface** coordinates and the sprite ignores camera position, zoom and shake — HUDs, on-screen buttons, overlays. Touch is mapped back automatically, so `tap` still works. Works on any sprite, not only text |
+| `scrollFactor` | `1` | Parallax: how much camera travel (and shake) moves this sprite. `0.5` = a background layer at half speed, `1.5` = a foreground layer overtaking the camera, `0` = pinned to the view but **still zooming** with `cameraScale` (unlike `screenFixed`). Rendering and touch mapping only — `x`/`y`, physics and collisions stay in plain world coordinates. Ignored when `screenFixed` is true |
 | `pixelSnap` | `false` | Rounds only the rendered anchor to a framebuffer pixel after camera position and zoom. Physics and live `x`/`y` stay subpixel floats. Combine with `smoothing: false` when a moving pixel-art sprite must keep a stable texel phase |
 
 ### Sheet and animation
@@ -214,19 +247,20 @@ Every sprite gets its own phase, and the wobble unwinds exactly when disabled. T
 | `glowColor` | — | Tinted, blurred silhouette drawn behind the sprite by a shader pass |
 | `glowBlur` | `0` | Blur radius in px; `0` = off. An active glow switches to the silhouette shader and back: **2 extra draw calls per glowing sprite per frame** (2 more while a `flash()` runs), even on a shared texture. Fine for a few highlights, not for every coin in the level |
 | `glowOpacity` | `1` | Halo strength 0..1 — tweenable via `animate`, so a glow can fade in without touching the blur |
-| `blend` | `'normal'` | `'add'` brightens the backdrop instead of covering it (glows, fire, lasers). Costs one batch flush per mode change — group additive sprites by `zIndex` |
+| `blend` | `'normal'` | `'add'` brightens the backdrop (glows, fire, lasers), `'multiply'` darkens it (contact shadows, stains, grime), `'screen'` lightens softly without blowing out to white (fog, god rays, soft light). Costs one batch flush per mode change — group same-blend sprites by `zIndex`. Unknown strings fall back to `'normal'` silently, and the names are case-sensitive |
 
 ### Methods
 
 | Method | Notes |
 | --- | --- |
-| `play(name)` | Start a named animation |
-| `stop()` | Stop the animation on the current frame |
+| `play(name, options)` | Start a named animation; returns `false` for an unknown name. `options.then` (a name or an array of names) chains natively — each queued animation starts as the previous non-looping one finishes, and `animationcomplete` still fires per step. A looping animation ends the chain |
+| `stop()` | Stop the animation on the current frame; also drops any queued `then` chain |
+| `followPath(points, options)` | Walk the sprite along `points` — `{ x, y }` objects or `[x, y]` pairs, **at least two**, fewer is ignored with a warning. `options`: `speed` px/s (default `100`), `loop` (closed circuit), `rotate` (face along the path, 0 = up), `smoothing` (corner radius in px, precomputed once). Fires `pathcomplete` at the end of a non-looping run and clears itself. `followPath(null)` stops in place |
 | `animate(options)` | Native tween, fires `complete` |
 | `clearTweens()` | Cancel tweens in progress — call before starting a replacement tween |
 | `flash(color, duration)` | Fills the sprite's silhouette with `color` (default white) and fades it out over `duration` ms (default 150). The classic damage/invincibility flash that a multiplicative `tintColor` cannot do |
 
-`animate()` accepts exactly these keys — anything else is silently ignored: `x`, `y`, `scale`, `scaleX`, `scaleY`, `rotation`, `opacity`, `glowOpacity`, `duration` (ms), `delay` (ms), `easing`, `frame` (the sheet frame to set once the tween finishes). Chain moves by re-calling `animate` from the `complete` handler — that is how patrols and ping-pong motions are built.
+`animate()` accepts exactly these keys — anything else is silently ignored: `x`, `y`, `scale`, `scaleX`, `scaleY`, `rotation`, `opacity`, `glowOpacity`, `duration` (ms), `delay` (ms), `easing`, `frame` (the sheet frame to set once the tween finishes). Re-calling `animate` from the `complete` handler is still how a ping-pong or a blink is built (there is no `repeat`/`yoyo` yet), but a fixed route is now `followPath` instead of a chain of tween legs.
 
 ### Sprite events
 
@@ -240,8 +274,9 @@ Every sprite gets its own phase, and the wobble unwinds exactly when disabled. T
 | `dragend` | `x`, `y` | Finger lifted; the sprite has already moved natively |
 | `pinch` | `scaleX`, `scaleY` | While two-finger scaling |
 | `rotate` | `rotation` | While two-finger rotating |
-| `animationcomplete` | `animation` | A non-looping sheet animation finished |
+| `animationcomplete` | `animation` | A non-looping sheet animation finished — including each finished step of a `then` chain |
 | `complete` | final transform values | A tween finished |
+| `pathcomplete` | `x`, `y` | A non-looping `followPath` run reached the end |
 | `collision` | `group`, `other`, `x`, `y` | Overlap with a `collidesWith` group began |
 | `collisionend` | `group`, `other`, `x`, `y` | That overlap ended: the shapes separated, **or** the partner was removed from the scene, hidden (`visible = false`) or stopped matching the group filter |
 | `land` | `x`, `y`, `other`, `group` | Landed on top of a `solidWith` solid. `other` is the solid — read it to react to what you landed on (trampolines, hazards) |
@@ -349,7 +384,7 @@ Add and remove like a sprite: `gameView.add(emitter)` / `remove(emitter)`. Spawn
 | `startScale` / `endScale` | `1` / `1` | Interpolated over `lifetime` |
 | `startOpacity` / `endOpacity` | `1` / `0` | Interpolated over `lifetime` |
 | `tint` | white | Tint white particle art at runtime — one tiny texture covers every color in the game |
-| `blend` | `'normal'` | `'add'` for fire, sparks and magic |
+| `blend` | `'normal'` | `'add'` for fire, sparks and magic; `'multiply'` for smoke and dust that darken what they cross; `'screen'` for soft haze |
 | `maxParticles` | `200` | Pool size, hard cap 1000 |
 
 Methods: `emit(n)` fires a one-shot burst on top of `rate`; `clear()` kills all live particles.
@@ -381,8 +416,8 @@ The tether yields at the end no finger owns: with a fixed head anchor the tail s
 
 Only these events exist. Nothing fires per frame, and events are only fired if a listener is registered.
 
-- **GameView**: `press`, `tap`, `release`, `resize`
-- **Sprite** (and therefore **Text**): `press`, `release`, `tap`, `dragstart`, `drag`, `dragend`, `pinch`, `rotate`, `animationcomplete`, `complete`, `collision`, `collisionend`, `land`
+- **GameView**: `press`, `tap`, `release`, `resize`, `timer`
+- **Sprite** (and therefore **Text**): `press`, `release`, `tap`, `dragstart`, `drag`, `dragend`, `pinch`, `rotate`, `animationcomplete`, `complete`, `pathcomplete`, `collision`, `collisionend`, `land`
 - **SpriteSheet, Font, Sound, Emitter, Rope**: none
 
 There is `collision` (enter) and `collisionend` (exit), but deliberately **no** per-frame "stay" event — that would be bridge traffic every frame. Hold the in-between state in JS: you heard the enter, you will hear the end.
@@ -392,7 +427,7 @@ There is `collision` (enter) and `collisionend` (exit), but deliberately **no**
 - **`maxSpeed` caps `thrust` and `carMode` only — not plain velocity.** The clamp runs inside the thrust integration and inside the car model, so a sprite given `velocityX = 3000` travels at 3000 no matter what `maxSpeed` says. A ship accelerating on `thrust`, on the other hand, stops gaining speed at 500 px/s unless you raise its own `maxSpeed`, which is the cap people actually hit.
 - **`sprite.animation` survives `stop()`.** Comparing it to decide whether to restart an animation gives a false negative. The top-down demo tracks a `walking` boolean itself.
 - **`follow()` resets its options.** See the GameView section.
-- **`animate()` ignores unknown keys.** Tweening `width`, `height` or `tintColor` silently does nothing — those are instant writes only.
+- **`animate()` ignores unknown keys.** Tweening `width`, `height` or `tintColor` silently does nothing — those are instant writes only. There is no `repeat` or `yoyo` either; re-launch from `complete`.
 - **`visible: false` removes a sprite from collision too.** That is why pooled bullets and rocks use it instead of removing them from the scene.
 - **`scaleX: -1` versus `flipX: true`.** Negative scale flips the physics and hit-test shape with the art; `flipX` mirrors only the drawing. Use `flipX` unless you specifically want the former.
 - **Tween positions are absolute.** A sprite with `idleAnimation` on will land off-target; disable the wobble before the tween and re-enable it on `complete`.
@@ -404,5 +439,18 @@ There is `collision` (enter) and `collisionend` (exit), but deliberately **no**
 - **A swept bullet can enter and leave in consecutive frames.** If it crosses a thin target entirely, you get `collision` and then `collisionend` almost immediately — react on the enter.
 - **Text `width`/`height` are read-only in practice.** They report the laid-out block; to make text bigger use `scale`.
 - **`hitboxShape: 'circle'` also changes the touch area**, not just collisions.
-- **`blend: 'add'` costs a batch flush per mode change.** Group additive sprites on their own `zIndex` band.
+- **Every blend-mode change costs a batch flush.** Group same-blend sprites on their own `zIndex` band; alternating modes sprite by sprite degrades toward one draw call each.
 - **`glowColor` alone draws nothing.** The glow pass only runs when `glowBlur > 0` *and* `glowOpacity > 0`; both default in a way that makes `glowColor` on its own a no-op (`glowBlur` is `0`). Always set the blur radius with the color.
+- **`followPath` overrides position absolutely.** The path is applied after velocity and gravity have been integrated and writes `x`/`y` outright, so a path-driven sprite ignores its own physics for placement. Do not expect `velocityX` and a path to add up. A tween on `x`/`y` runs *after* the path in the same frame and wins outright — pick one.
+- **A path needs at least two points.** One point (or a non-array) clears the path and logs a warning instead of parking the sprite there.
+- **Path movement still carries riders.** It feeds the frame delta like a tweened or velocity-driven solid, so a platform on a `followPath` circuit moves whatever stands on it.
+- **`raycast` groups: pass an array, not varargs.** Android accepts both `raycast(x0, y0, x1, y1, ['enemy'])` and loose arguments; iOS reads only the array. Write the array and it works on both.
+- **`raycast` only sees `visible`, non-`screenFixed` sprites carrying a `collisionGroup`.** A hidden pooled sprite is invisible to the ray, which is usually what you want; an untagged wall is invisible too, which usually is not. HUD sprites are skipped by design, so a screen-fixed score never blocks a world-space line of sight.
+- **`raycast` is a discrete query, not a sensor.** Calling it from a game-clock timer or a tap handler is the intended use. Calling it every frame from JS puts exactly the bridge traffic in the loop that the whole engine exists to avoid.
+- **`findPath` walks the sprite's center, not its silhouette.** The waypoints are a line through free grid cells, so a body wider than a cell clips corners unless you pass `clearance` (roughly half the walker's width) or shrink `cellSize`. Obstacles are read from the **hitbox**, so `hitboxScale` and `hitboxShape: 'circle'` shape the walkable space too.
+- **`findPath` returning `null` is not always "no route".** Degenerate `bounds`, `cellSize <= 0` and a grid over ~1M cells all return `null` as well — a `bounds` rect of the whole world at `cellSize: 4` fails silently that way. Check the numbers before blaming the maze.
+- **A path of one or two points is the straight-line case, not a bug.** With nothing in the way `findPath` returns just `[start, goal]`; `path.length < 2` is the guard the demos use to detect a tap outside `bounds`.
+- **Game-clock timers are not `setTimeout`.** `after`/`every` freeze at `timeScale: 0` and pause with the render loop — which is why spawn waves belong there — but a pause menu that needs a real countdown still needs `setTimeout`.
+- **`screenFixed` beats `scrollFactor`.** Setting both leaves the sprite screen-fixed; the parallax offset is skipped entirely.
+- **Ghost lines along a frame's edge are a filtering artifact, not bad art.** A magnified `smoothing: true` sheet samples past the frame border — 1px seams, or the next row's heads at the bottom. Grid sheets have inset UVs upstream since 2026-08-19; an atlas needs padding and extrude at pack time, and an older module build needs either `smoothing: false` or a rebuild.
+- **`scrollFactor` does not move the sprite.** It shifts where the sprite is *drawn* and maps touch back to match. `x`, `y`, physics and collisions stay in world coordinates, so a `scrollFactor: 0.5` platform still collides where its world `x` says, not where it looks.

package/skills/ti-game/references/project-setup.md

@@ -22,8 +22,8 @@ There is no registry and no `ti module install` command — Titanium modules are
 
 ```
 modules/
-├── android/ti.game/0.3.0/
-└── iphone/ti.game/0.3.0/
+├── android/ti.game/0.4.0/
+└── iphone/ti.game/0.4.0/
 ```
 
 Unzipping the module zip at the project root creates exactly that layout.
@@ -34,14 +34,14 @@ Then declare it in `tiapp.xml`. The iOS platform key is `iphone`, not `ios`:
 
 ```xml
 <modules>
-  <module platform="android" version="0.3.0">ti.game</module>
-  <module platform="iphone" version="0.3.0">ti.game</module>
+  <module platform="android" version="0.4.0">ti.game</module>
+  <module platform="iphone" version="0.4.0">ti.game</module>
 </modules>
 ```
 
 Pinning `version` is worth the extra attribute: with two module versions installed side by side, an unpinned entry silently picks the highest one.
 
-The version number lies right now: the manifest has read `0.3.0` through several feature landings. A zip built before 2026-08-18 has no `createText`, no `createFont`, no `screenFixed`, no `swept` and no `collisionend` while still calling itself 0.3.0. If a call that this skill documents is `undefined` at runtime, rebuild the module from upstream `main` rather than hunting for a typo.
+The manifest went `0.3.0` → **`0.4.0`** on 2026-08-20, after months of feature landings under the old number — so `0.4.0` is the first version string that actually means something: it carries everything this skill documents. `0.3.0` does not, and there are several different `0.3.0` builds in the wild. One built before 2026-08-18 has no `createText`, no `createFont`, no `screenFixed`, no `swept` and no `collisionend`; one built before 2026-08-19 also lacks `followPath`, `play(name, { then })`, `raycast`, `after`/`every`, `scrollFactor` and the `multiply`/`screen` blend modes; one built before 2026-08-19 16:00 UTC+2 additionally bleeds at grid-frame edges on smoothed sheets. `findPath` exists only in `0.4.0`. If a call that this skill documents is `undefined` at runtime, check the manifest version first and rebuild from upstream `main` rather than hunting for a typo.
 
 Building the module from source:
 
@@ -53,7 +53,7 @@ ti build -p ios --build-only        # macOS only → ios/dist/ti.game-iphone-<ve
 
 ## Classic projects
 
-`require('ti.game')` anywhere and add the GameView to a window. The 21 demos in the module's `example/` folder are Classic, each one a CommonJS module exporting a start function:
+`require('ti.game')` anywhere and add the GameView to a window. The 24 demos in the module's `example/` folder are Classic, each one a CommonJS module exporting a start function:
 
 ```javascript
 // Resources/flappy.js
@@ -218,6 +218,8 @@ Practical rules:
 
 - **One sheet per scene if you can manage it.** Sprites sharing a texture batch into a single draw call. The endless-runner demo even draws its crash burst from the *player's* sheet to keep the whole scene on one texture.
 - **`smoothing: false` for pixel art.** Combine with `pixelSnap: true` on moving sprites when the texel phase must stay stable.
+- **Ship sheets as PNG, not JPG.** JPEG has no alpha channel, so every frame comes back on an opaque block, and its compression smears colour across frame borders — exactly where a sheet can least afford it. The module's own `basic.js` demo was converted from `hero.jpg` to `hero.png` on 2026-08-19 for this reason.
+- **Edge bleeding on a smoothed grid is fixed upstream, on grids only.** Linear filtering samples past a frame's edge when the sprite is magnified, which shows up as 1px ghost lines or the next row's heads peeking in at the bottom. Since 2026-08-19 a grid sheet with `smoothing: true` pulls its *interior* frame edges in by half a texel automatically; outer edges stay at the texture border, so full-texture `tileRepeat` frames still wrap seamlessly, and `smoothing: false` sheets skip the inset entirely (NEAREST cannot bleed, and pixel art at 1:1 needs exact UVs). **Atlas sheets are not covered** — their UVs come straight from the JSON, so prevent bleeding when packing instead: 2px padding and extrude in TexturePacker.
 - **`repeat: true` for tiling.** Required by sprites using `tileRepeat`, and it needs power-of-two texture dimensions on ES 2.0 — a 512×64 street strip tiles, a 500×60 one does not.
 - **White art plus `tint`** covers every color variant of a particle or effect with one small texture.
 

package/skills/ti-game/references/recipes.md

@@ -1,13 +1,13 @@
 # ti.game recipes by genre
 
-Working patterns distilled from the 21 official demos in `example/` and from a shipped Alloy game (Titanium Lander). Each recipe shows the part that is specific to the genre; the shared scaffolding is in [The scaffolding](#the-scaffolding) and is not repeated.
+Working patterns distilled from the 25 official demos in `example/` and from a shipped Alloy game (Titanium Lander). Each recipe shows the part that is specific to the genre; the shared scaffolding is in [The scaffolding](#the-scaffolding) and is not repeated.
 
-Every snippet uses only API that exists in upstream `main` as of 2026-08-18. Check [roadmap.md](roadmap.md) before reaching for anything not shown here.
+Every snippet uses only API that exists in upstream `main` as of 2026-08-20 (manifest `0.4.0`). Check [roadmap.md](roadmap.md) before reaching for anything not shown here.
 
 ## Contents
 
 - [The scaffolding](#the-scaffolding)
-- [Cross-cutting patterns](#cross-cutting-patterns) — pooling, parallax, multitouch controls, HUD text, invisible triggers, enter/exit zones, hit-stop
+- [Cross-cutting patterns](#cross-cutting-patterns) — pooling, parallax, multitouch controls, HUD text, invisible triggers, enter/exit zones, patrol routes, line of sight, pathfinding, game-clock timers, hit-stop
 - [Tap-to-flap](#tap-to-flap) (`flappy.js`)
 - [Platformer](#platformer) (`platformer.js`)
 - [Top-down / Zelda](#top-down--zelda) (`topdown.js`)
@@ -18,11 +18,11 @@ Every snippet uses only API that exists in upstream `main` as of 2026-08-18. Che
 - [Rhythm game](#rhythm-game) (`rhythm.js`)
 - [Ball sports / breakout physics](#ball-sports--breakout-physics) (`volley.js`)
 - [Cards and board games](#cards-and-board-games) (`cards.js`, `puzzle.js`)
-- [Point & click adventure](#point--click-adventure) (`pointclick.js`)
+- [Point & click adventure](#point--click-adventure) (`pointclick.js`, `maze.js`)
 - [Particles](#particles) (`particles.js`)
 - [Ropes and chains](#ropes-and-chains) (`rope.js`)
 - [Camera work](#camera-work) (`camera.js`)
-- [Visual effects](#visual-effects) (`blend.js`, `flip.js`, `timescale.js`)
+- [Visual effects](#visual-effects) (`blend.js`, `flip.js`, `timescale.js`, `demoscene.js`)
 
 ## The scaffolding
 
@@ -139,9 +139,21 @@ The `generation` counter matters: without it, a timeout fired for an old shot de
 
 Recycling a sprite with `visible = false` fires `collisionend` on anything it was touching. Handlers that treat separation as "the player walked away" need to tolerate a despawn.
 
-### Seamless parallax
+### Parallax
 
-Two screen-wide copies per layer, scrolled by native velocity. `wrapX`/`wrapShift` teleport a copy that left the screen back behind its sibling. No JS runs in the loop.
+Two ways, and the choice is decided by whether the **camera** moves or the **world** does.
+
+**A camera travelling through a static world** — platformer, top-down, anything using `follow()` — wants `scrollFactor`, one property per layer:
+
+```javascript
+const clouds = Game.createSprite({ sheet: cloudSheet, x: W / 2, y: 120, zIndex: 0, scrollFactor: 0.25 });
+const hills  = Game.createSprite({ sheet: hillSheet,  x: W / 2, y: 260, zIndex: 1, scrollFactor: 0.6 });
+const sun    = Game.createSprite({ sheet: sunSheet,   x: 60,    y: 60,  zIndex: 0, scrollFactor: 0 });
+```
+
+`0` pins the sun to the view while still zooming with `cameraScale` — that is the difference from `screenFixed`, which ignores zoom as well. Values above `1` overtake the camera for foreground grass. The sprite's `x`/`y` never change: `scrollFactor` shifts where it is *drawn* and maps touch back to match, so collisions still happen at the world coordinates you wrote.
+
+**A world scrolling under a fixed camera** — endless runner, flappy — cannot use `scrollFactor`, because the camera never travels. Two screen-wide copies per layer, scrolled by native velocity; `wrapX`/`wrapShift` teleport a copy that left the screen back behind its sibling. No JS runs in the loop.
 
 ```javascript
 function makeLayer(sheet, y, height, speed, z) {
@@ -172,7 +184,7 @@ const skyline = makeLayer(skylineSheet, groundTop - SKY_H / 2, SKY_H, SPEED * 0.
 const street = makeLayer(streetSheet, (groundTop + H) / 2, H - groundTop, SPEED, 8);
 ```
 
-Slower layers read as further away. Three layers (clouds at 0.15×, hills at 0.4×, ground at 1×) is the flappy demo's recipe.
+Slower layers read as further away. Three layers (clouds at 0.15×, hills at 0.4×, ground at 1×) is the flappy demo's recipe. `camera.js` demos the other half: `scrollFactor` cloud shadows at 1.35× and a `scrollFactor: 0` sun.
 
 ### Multitouch on-screen controls
 
@@ -317,6 +329,142 @@ The listener goes on the sprite that declares `collidesWith`. For a pressure pla
 
 Removing the partner from the scene, hiding it (`visible = false`) or changing its `collisionGroup` mid-contact all count as separation, so the exit still fires and the door cannot stick open when an object despawns.
 
+### Patrol routes and animation chains — `path.js`
+
+A fixed route is `followPath`, not a chain of tween legs. The points are walked natively at `speed` px/s; `rotate: true` turns the sprite to face along the path (heading `0` = up, same convention as `rotation`), and `smoothing` rounds the corners with a radius in px.
+
+```javascript
+guard.followPath([
+	{ x: W * 0.22, y: H * 0.46 },
+	{ x: W * 0.78, y: H * 0.46 },
+	{ x: W * 0.78, y: H * 0.60 },
+	{ x: W * 0.22, y: H * 0.60 }
+], { speed: W * 0.16, loop: true });     // sharp rectangle, walks forever
+
+ship.followPath(diamond, {
+	speed: W * 0.35,
+	loop: true,
+	rotate: true,                          // nose follows the heading
+	smoothing: W * 0.12                    // rounds the diamond into an oval
+});
+```
+
+A non-looping run fires `pathcomplete` and clears itself; `followPath(null)` stops in place. Two points is the minimum — one point is ignored with a warning, not treated as "go there".
+
+The path writes `x`/`y` outright, after velocity and gravity have been integrated, so **a sprite on a path ignores its own physics for placement**. It still feeds the frame delta, though, which means a platform on a looping circuit carries whatever is standing on it — a moving platform is now a path, not a tween ping-pong.
+
+Animation chains are native in the same spirit:
+
+```javascript
+guard.play('hop', { then: 'walk' });        // one call, no animationcomplete handler
+bird.play('flap', { then: ['glide', 'idle'] });
+```
+
+Each queued animation starts as the previous non-looping one finishes, and a looping animation ends the chain. `animationcomplete` still fires per step if you want to hook one. `stop()` drops the queue along with the current animation.
+
+### Line of sight, ledge probes and hitscan — `raycast.js`
+
+`gameView.raycast(x0, y0, x1, y1, groups)` answers "what is the first thing on this line?" with `null` or `{ x, y, distance, group, sprite, normal }`. The targets need nothing but a `collisionGroup` — no `collidesWith` on the asking side.
+
+```javascript
+const hit = gameView.raycast(eyeX, eyeY, dog.x, dog.y, ['blocker']);
+if (hit) {
+	aimBeam(sight, eyeX, eyeY, hit.x, hit.y, hit.distance);
+	sight.tintColor = '#f44';               // view blocked at the impact point
+}
+```
+
+Pass the groups as an **array**. Android also accepts loose arguments, iOS does not, so the array is the portable form. Only `visible`, non-`screenFixed` sprites carrying a `collisionGroup` are candidates — a pooled sprite parked at `visible = false` is correctly invisible to the ray, an untagged wall is invisible too (the usual mistake), and HUD sprites are skipped by design so a screen-fixed score never blocks a world-space sight line.
+
+A ledge probe is the same call pointed straight down, one step ahead of the feet:
+
+```javascript
+const dir = walker.velocityX >= 0 ? 1 : -1;
+const probeX = walker.x + dir * W * 0.07;
+if (!gameView.raycast(probeX, walker.y, probeX, walker.y + W * 0.12, ['ground'])) {
+	walker.velocityX = -walker.velocityX;   // nothing underfoot: turn before falling
+	walker.flipX = walker.velocityX < 0;
+}
+```
+
+Run these from a coarse timer or a tap handler — the demo probes at 150 ms — never once per frame. A per-frame `raycast` from JS is exactly the bridge traffic the engine is built to avoid. Note that `raycast.js` drives its probes with `setInterval`: it was written two commits before game-clock timers existed, and `gameView.every` is the better home for an AI tick today.
+
+### Pathfinding around obstacles — `maze.js`, `pointclick.js`
+
+`gameView.findPath(from, to, options)` is grid A\* over the same sprites `raycast` sees: visible, tagged with a `collisionGroup`, nothing else to set up. It returns `{ x, y }` waypoints — or `null` — and those waypoints go straight into `followPath`, so the walk itself runs natively.
+
+```javascript
+gameView.addEventListener('tap', (e) => {
+	const path = gameView.findPath(
+		{ x: player.x, y: player.y },
+		{ x: e.x, y: e.y },
+		{ cellSize: TILE, groups: ['wall'], bounds: MAZE_BOUNDS });
+	if (!path || path.length < 2) {
+		return;                                   // tapped outside the walkable rect
+	}
+	player.scaleX = path[1].x < player.x ? -1 : 1;   // face the first leg
+	player.play('walk');
+	player.followPath(path, { speed: SPEED });
+});
+```
+
+`cellSize` is the one option that decides everything else. On a tile map, pass the tile size and the grid lines up with the walls exactly; on a free-form scene, something near the walker's width. Too coarse and doorways disappear, too fine and the query gets expensive (over ~1M cells it returns `null` rather than allocating).
+
+**The path is a line for the sprite's *center*.** A body wider than a cell scrapes corners unless you pass `clearance` — about half the walker's width:
+
+```javascript
+const path = gameView.findPath(from, to, {
+	cellSize: Math.round(W * 0.04),
+	groups: ['obstacle'],
+	clearance: PLAYER_W * 0.35,               // keep half a body off the trunk
+	bounds: WALK_BOUNDS                       // the walkable floor, in center coordinates
+});
+```
+
+`bounds` doubles as the walkable area, which is how the point-&-click floor clamp survives: `WALK_BOUNDS` is the old `WALK_TOP`/margin clamp expressed as a rect, in **sprite-center coordinates** (tap points are feet positions, so subtract half the sprite height before searching).
+
+**Obstacles do not have to be the art.** `pointclick.js` blocks only the oak's trunk with an invisible sprite — `opacity: 0`, `touchEnabled: false`, no collision wiring — because `findPath` reads the hitbox, not the pixels. The canopy stays walkable, so the player still passes behind the tree with `ySort` doing the depth:
+
+```javascript
+gameView.add(Game.createSprite({
+	sheet: treeSheet, opacity: 0, touchEnabled: false,
+	x: tree.x, y: groundY, width: TREE_W * 0.35, height: PLAYER_H * 0.8,
+	collisionGroup: 'obstacle'
+}));
+```
+
+A blocked goal snaps to the nearest free cell a few rings out, so tapping a wall walks to its edge instead of doing nothing. Tapping outside `bounds` is the case worth guarding — that is what `path.length < 2` catches.
+
+**Chasing AI is a timer, not a frame loop.** Re-path on a coarse game-clock tick and let `followPath` cover the gap:
+
+```javascript
+const chase = gameView.every(800, () => {
+	const path = gameView.findPath({ x: hound.x, y: hound.y }, { x: player.x, y: player.y }, PATH_OPTIONS);
+	if (path && path.length > 1) {
+		hound.followPath(path, { speed: TILE * 2.6 });
+	}
+});
+win.addEventListener('close', () => gameView.cancelTimer(chase));
+```
+
+`followPath(null)` cancels an in-flight route — `maze.js` uses it to yank the hound back to its den after it catches you.
+
+To *see* what A\* did, ask twice: `simplify: false` returns every grid cell it walked, `simplify: true` (the default) the line-of-sight-reduced corners the sprite actually gets. Dropping a dot on each is the whole route visualization in the maze demo.
+
+### Timers on the game clock
+
+`gameView.after(ms, cb)` and `every(ms, cb)` run on the engine's clock: they stretch under `timeScale` slow motion, freeze at `0`, and pause with the render loop. Spawn waves, AI decision ticks and respawn delays belong here — a paused game that keeps spawning is the bug this fixes.
+
+```javascript
+const wave = gameView.every(2000, () => spawnEnemy());
+gameView.after(500, () => player.play('idle'));
+gameView.cancelTimer(wave);                 // both return an int id
+```
+
+The callback receives `{ id }`. Called without a callback, the view fires a `timer` event carrying the same id instead — pick one, not both. A repeating timer fires at most once per frame and restarts its interval after a stall rather than bursting to catch up.
+
+What still belongs in `setTimeout`: anything that must keep running while the game is frozen. A pause menu counting down in real seconds is not on the game clock.
+
 ### Hit-stop on impact
 
 Freezing the scene for a few frames makes an impact read before the explosion starts. `timeScale = 0` stops the engine from advancing while it keeps drawing and accepting touches — no render loop to restart, no state to save. (Pattern from Titanium Lander: 140 ms on a crash, 90 ms when a shield absorbs a hit.)
@@ -1049,31 +1197,37 @@ Because every handler is per sprite, this is multi-touch for free: each finger g
 
 ## Point & click adventure
 
-Tap-to-walk with a distance-sized tween gives constant walking speed regardless of distance:
+Tap-to-walk routes around the scenery with `findPath` and walks the waypoints natively:
 
 ```javascript
+const WALK_BOUNDS = {                                     // the floor, in sprite-CENTER coordinates
+	minX: W * 0.06, maxX: W * 0.94,
+	minY: WALK_TOP - PLAYER_H / 2, maxY: H * 0.95 - PLAYER_H / 2
+};
+
 function walkTo(x, y) {
-	const targetX = Math.min(Math.max(x, W * 0.06), W * 0.94);
-	const targetY = Math.min(Math.max(y, WALK_TOP), H * 0.95) - PLAYER_H / 2;
-	const distance = Math.sqrt(Math.pow(targetX - player.x, 2) + Math.pow(targetY - player.y, 2));
-	if (distance < 4) {
-		return;
+	const targetX = Math.min(Math.max(x, WALK_BOUNDS.minX), WALK_BOUNDS.maxX);
+	const targetY = Math.min(Math.max(y - PLAYER_H / 2, WALK_BOUNDS.minY), WALK_BOUNDS.maxY);
+	const path = gameView.findPath({ x: player.x, y: player.y }, { x: targetX, y: targetY }, {
+		cellSize: Math.round(W * 0.04),
+		groups: ['obstacle'],
+		clearance: PLAYER_W * 0.35,                       // keep half a body off the trunk
+		bounds: WALK_BOUNDS
+	});
+	if (!path || path.length < 2) {
+		return;                                           // tapped outside the floor
 	}
-	player.scaleX = targetX < player.x ? -1 : 1;
-	player.clearTweens();                                 // cancel a walk in progress
+	player.scaleX = path[1].x < player.x ? -1 : 1;        // face the first leg
 	player.play('walk');
-	player.animate({
-		x: targetX,
-		y: targetY,
-		duration: distance / WALK_SPEED * 1000,           // constant px/s
-		easing: Game.EASE_LINEAR                          // no ease on a walk
-	});
+	player.followPath(path, { speed: WALK_SPEED });       // constant px/s, cancels the previous route
 }
 
-player.addEventListener('complete', () => player.play('idle'));
+player.addEventListener('pathcomplete', () => player.play('idle'));
 ```
 
-The walkable floor is a clamp (`WALK_TOP`), not geometry — no collision needed for a classic adventure floor.
+The walkable floor is still a clamp, not geometry — but it is now expressed as the `bounds` rect the search runs in. Obstacles are separate: `pointclick.js` adds an invisible box over the oak's *trunk* only (`opacity: 0`, `touchEnabled: false`, `collisionGroup: 'obstacle'`), so the player circles the trunk and still walks behind the canopy with `ySort`. See [Pathfinding around obstacles](#pathfinding-around-obstacles--mazejs-pointclickjs) for the options.
+
+With nothing in the scene to route around, `findPath` returns `[start, goal]` and `followPath` walks the straight line — the same result as the distance-sized `animate()` tween this demo used before, with one less speed calculation to get wrong.
 
 **Hotspots and the tap conflict.** The GameView fires `tap` for *every* touch, including ones a sprite already handled, so a naive walk handler also walks when you click the bird. Hit-test the interactive sprites in JS and bail out:
 
@@ -1229,7 +1383,16 @@ For a big ground plane, one `tileRepeat` sprite covering the whole world beats a
 
 ## Visual effects
 
-**`blend: 'add'`** — identical sprites, different blend: normal overlaps cover each other, additive overlaps sum and bloom toward white. Use it for glows, fire, lasers and light shafts, and group additive sprites on their own `zIndex` band since each blend change costs a batch flush.
+**Blend modes** — four, on sprites and emitters alike. `blend.js` shows all of them side by side over a bright meadow strip:
+
+| Mode | Does | Use for |
+| --- | --- | --- |
+| `'normal'` | Covers what is behind it | Everything else |
+| `'add'` | Sums toward white | Glows, fire, lasers, light shafts, sparks |
+| `'multiply'` | Darkens the backdrop | Contact shadows under sprites, stains, grime, smoke that dims |
+| `'screen'` | Lightens without blowing out | Fog, haze, god rays, soft light |
+
+Group same-blend sprites on their own `zIndex` band: every mode change in draw order costs a batch flush, so alternating them sprite by sprite degrades toward one draw call each. Unknown strings fall back to `'normal'` in silence and the names are case-sensitive — `blend: 'Add'` renders as normal with no warning.
 
 **`flash(color, duration)`** — a solid-color silhouette that fades out natively. This is the damage flash a multiplicative `tintColor` cannot produce (white tint = no change). Retrigger it on a timer for invincibility blinking.
 
@@ -1263,6 +1426,6 @@ gameView.addEventListener('tap', () => {
 });
 ```
 
-**`timeScale`** — one property slows or freezes physics, sheet animations, tweens, particles and the camera together, while rendering and touch keep running. `0.5` for slow motion, `0` for a pause that still draws, short bursts of `0` for hit-stop.
+**`timeScale`** — one property slows or freezes physics, sheet animations, tweens, particles, the camera and the game-clock timers together, while rendering and touch keep running. `0.5` for slow motion, `0` for a pause that still draws, short bursts of `0` for hit-stop. `timescale.js` makes the timer half visible: a clock on `gameView.every(1000, ...)` freezes with the scene while a `setInterval` clock beside it keeps ticking.
 
 **`idleAnimation`** — a per-sprite organic sway that composes with tweens and drags. Cheap life for cards, icons, floating pickups and menu art. Disable it before a tween that must land precisely.

package/skills/ti-game/references/roadmap.md

@@ -1,6 +1,6 @@
 # What does not exist yet — and what to do instead
 
-The most expensive mistake with `ti.game` is writing code against a feature that sounds like it should be there because Phaser, Godot or Cocos has it. This file is the guard rail: everything below is **absent from upstream `main` as of 2026-08-18**, with the workaround that is idiomatic today.
+The most expensive mistake with `ti.game` is writing code against a feature that sounds like it should be there because Phaser, Godot or Cocos has it. This file is the guard rail: everything below is **absent from upstream `main` as of 2026-08-20** (manifest `0.4.0`), with the workaround that is idiomatic today.
 
 Planned items come from the module's `TODO.md`. Planned is not shipped — when a game needs one of these, use the workaround and keep the code isolated enough to swap later.
 
@@ -8,22 +8,22 @@ Planned items come from the module's `TODO.md`. Planned is not shipped — when
 
 | You might reach for | Status | What to do now |
 | --- | --- | --- |
-| A tilemap layer, Tiled JSON import, collision layer | Planned (priority 4) | One sprite per tile. Fine at 12×15, dead at 200×200. For large ground planes use a single `tileRepeat` sprite instead of a grid, and add sparse invisible solids for collision |
+| A tilemap layer, Tiled JSON import, collision layer | Planned (priority 4) | One sprite per tile. Fine at 12×15, dead at 200×200. For large ground planes use a single `tileRepeat` sprite instead of a grid, and add sparse invisible solids for collision. `findPath` already rasterizes those per-tile sprites, so a tile maze pathfinds today — the planned collision layer is meant to feed the same grid |
 | `sprite.parent`, attaching a turret to a tank, a hat to a hero, multi-part bosses | Planned (priority 7), the biggest structural item | Move the parts together from JS on the same coarse tick, or bake the combination into sheet frames. There is no transform inheritance |
-| `gameView.raycast(...)`, line-of-sight, ground probes | Planned (priority 3) | Compare positions in JS on a decision timer, or place an invisible trigger sprite where the probe would land and listen for `collision` |
+| `gameView.panTo(x, y, { duration, easing })` for a cutscene camera move | Planned (priority 1) | `follow()` an invisible sprite and tween that sprite, then `follow()` the player again |
+| `maxWidth` word wrap on text sprites | Planned (priority 5) | Break the lines yourself with `\n`. Measure with the font's `charWidth` on a monospace grid font; a proportional font needs a per-glyph tally |
+| Ping-pong animation playback, a per-sprite animation speed multiplier, a random start offset | Planned (priority 9) | Author the ping-pong into the sheet as a longer animation. Desync a field of torches by giving each one its own `idleAnimation` timing, or start them from staggered timers |
 | Native virtual joystick or d-pad bound to a sprite | Planned (priority 6) | Overlaid Titanium views, one per button, writing `velocityX`/`steering`/`throttle` on `touchstart`/`touchend`. Sibling views get simultaneous pointers, so multitouch works |
 | Gamepad support | Planned (priority 6) | Nothing native. Touch controls only |
-| `blend: 'multiply'` or `'screen'` | Planned (priority 2) | Only `'normal'` and `'add'` exist. For darkening, use `tintColor` (multiplicative) on the sprite itself |
 | Slopes in platformer terrain | Planned (priority 3), conditional | Build slopes as stepped rectangles, or keep terrain flat |
 | Per-frame animation events (footstep on frame 3) | Planned (priority 9) | Use `animationcomplete` on a short non-looping animation, or a timer aligned to the animation's fps |
-| `play('attack', { then: 'idle' })` chaining | Planned (priority 9) | Handle `animationcomplete` and call `play()` from it |
-| `sprite.followPath([points], ...)` | Planned (priority 9) | Chain tweens: call `animate()` again from the `complete` handler, one leg per call |
 | `playbackRate`, pitch jitter, stereo pan, `fadeTo()` on sounds | Planned (priority 8) | Only `volume`, `loop`, `play`, `pause`, `stop` exist. Ship a few pre-pitched variants of a sample and alternate between them |
 | An fps / draw-call / sprite-count overlay | Planned (developer experience) | `debug: true` draws collision shapes only. There is no stats overlay and no `performance` event — measure with platform tooling |
 | TypeScript definitions (`ti.game.d.ts`) | Planned (developer experience) | None yet |
 | Aseprite JSON import | Planned (developer experience) | TexturePacker JSON (hash or array) is the only atlas format |
 | A `stop()` for the whole scene | Not planned — already covered | `gameView.timeScale = 0` freezes everything the engine ticks while rendering and touch keep running. `gameView.pause()` stops the render loop entirely |
-| Tweening `width`, `height`, `tintColor` | Not planned | `animate()` only handles `x`, `y`, `scale`, `scaleX`, `scaleY`, `rotation`, `opacity`, `glowOpacity`, plus `frame` at the end. Everything else is an instant write |
+| Tweening `width`, `height` | Not planned | `animate()` only handles `x`, `y`, `scale`, `scaleX`, `scaleY`, `rotation`, `opacity`, `glowOpacity`, plus `frame` at the end. Everything else is an instant write — scale instead of sizing |
+| Tweening `tintColor`, or `repeat`/`yoyo` on `animate()` | Planned (priority 9) | Re-launch the tween from its `complete` handler for a blink or a ping-pong, and step a tint by writing `tintColor` on a game-clock timer |
 | Standard Titanium touch events on the GameView | By design | The view fires its own `press`, `tap`, `release` with scene coordinates. Sprites fire the richer set. Overlaid Titanium views keep their normal events |
 | A per-frame event to hook your own logic into | By design, never | The whole architecture exists to keep JS out of the frame loop. Use `collision`, `land`, `complete`, `animationcomplete`, or a coarse timer for decisions |
 
@@ -31,7 +31,7 @@ Planned items come from the module's `TODO.md`. Planned is not shipped — when
 
 The module's own `TODO.md` rules these out, each being a project the size of everything built so far:
 
-- **2D lighting and shadows.** If it ever comes up: a screen-dim overlay plus additive light sprites (`blend: 'add'`) gets 90% of the look for 1% of the cost.
+- **2D lighting and shadows.** If it ever comes up: a screen-dim overlay plus additive light sprites (`blend: 'add'`) gets 90% of the look for 1% of the cost, and `blend: 'multiply'` now does contact shadows under sprites.
 - **Skeletal animation (Spine).** Frame animations from sheets only.
 - **Full rigid-body physics (Box2D class).** What exists — velocity, gravity, solids with penetration resolution, restitution, circle and rect hitboxes, the car model, thrust — covers the target genres. Joints, torque, compound bodies and stacking do not exist.
 - **Render-to-texture.** The only offscreen pass is the fullscreen `cameraEffect`.
@@ -51,7 +51,13 @@ Recent additions that are easy to miss and easy to reimplement badly in JS:
 - **`screenFixed`** on any sprite: surface coordinates, camera position/zoom/shake ignored, touch mapped back. HUDs no longer need overlay views.
 - **`swept: true`** per sprite: the frame's movement is path-tested (Minkowski + slab) for both collision events and solid blocking. Fast bullets stop tunneling; `hitboxScale` tricks along the travel axis are obsolete.
 - **`collisionend`**: the exit half of the trigger lifecycle, including when the partner is removed, hidden or re-tagged mid-contact.
+- **`followPath(points, opts)` and `play(name, { then })`** (2026-08-19): patrol routes, bullet arcs and animation chains run natively. Do not chain `animate()` legs from `complete`, and do not re-`play()` from an `animationcomplete` handler.
+- **`gameView.raycast(x0, y0, x1, y1, groups)`** (2026-08-19): nearest hit as `{ x, y, distance, group, sprite, normal }`, or `null`. Line of sight, ledge probes and hitscan no longer need an invisible probe sprite — but it is a discrete query, not something to poll every frame.
+- **Game-clock timers** (2026-08-19): `gameView.after(ms, cb)` / `every(ms, cb)` / `cancelTimer(id)` obey `timeScale` and pause with the render loop. Spawn waves and AI ticks belong here, not in `setTimeout`.
+- **`scrollFactor`** (2026-08-19): per-sprite parallax against camera travel. A background layer is one property, not a hand-scrolled pair of copies.
+- **`blend: 'multiply'` and `'screen'`** (2026-08-19) join `'normal'` and `'add'`, on sprites and emitters alike.
+- **`gameView.findPath(from, to, options)`** (2026-08-20): grid A* around the sprites carrying a `collisionGroup`, returning waypoints ready for `followPath` — the Godot `AStar2D` / GameMaker `mp_grid` equivalent. Tap-to-walk and chasing AI route around obstacles natively now; hand-authored waypoint arrays and straight-line walking are the old workaround. Discrete like `raycast`: taps and AI timers, not per frame.
 
 ## Checking the current state
 
-This file describes upstream `main` on 2026-08-18 — a moving target: four items on this list shipped within two days of the skill being written, and the manifest version (`0.3.0`) did not change with them. Before relying on any "planned" item being still absent — or on a workaround still being necessary — read `TODO.md` and `README.md` at https://github.com/m1ga/ti.game, the module's upstream repo. `README.md` there is the canonical API documentation; the `documentation/` folder only points at it. The priorities quoted above are the maintainer's, not a roadmap this skill decides.
+This file describes upstream `main` on 2026-08-20 — a moving target: nine items on this list shipped within four days of the skill being written, four of them in a single day, and the manifest only caught up at the end (`0.3.0` → `0.4.0` on 2026-08-20, after all of them). Before relying on any "planned" item being still absent — or on a workaround still being necessary — read `TODO.md` and `README.md` at https://github.com/m1ga/ti.game, the module's upstream repo. `README.md` there is the canonical API documentation; the `documentation/` folder only points at it. The priorities quoted above are the maintainer's, not a roadmap this skill decides.