vibeview 0.25.0 → 0.27.0

package/skills/vibeview-agent/SKILL.md

@@ -38,10 +38,17 @@ https://vibeview.io/docs/preparing-your-build.
    ```
 2. Start Metro in the React Native project (`yarn start` or equivalent) and
    leave it running.
-3. Start a dev session in the background and get structured output:
+3. Start a dev session in the background and get structured output. No
+   `vibeview.json` yet in this project (first run here)? Pass `--platform`
+   and `--app` explicitly — non-interactive runs never prompt, so both must
+   be given up front (`--app` is `app_id` from step 1's upload output):
    ```bash
-   vibeview dev --detach --json
+   vibeview dev --platform ios --app <app_id> --detach --json
    ```
+   Once `vibeview.json` exists (this run or an earlier one), a bare
+   `vibeview dev --detach --json` reruns the same platform/app. A
+   `{"event":"error","code":"config_required",...}` line names exactly what's
+   still missing (usually `--platform` or `--app`) — pass it and retry.
 4. Parse the `session_ready` line from stdout:
    ```json
    {"event":"session_ready","session_id":"...","url":"...","pid":1234}
@@ -127,6 +134,9 @@ it rather than letting it appear unexplained in their `git status`.
   ```bash
   vibeview scroll-to "Delete account" --session <id>
   ```
+  A `scroll-to` that never finds its text, or a `type` whose field does not
+  end up holding the text, prints `error` and exits 1 — check the reason
+  line and the screen before retrying.
 - **Precise two-point gestures** (sliders, drag-and-drop reorder, map
   panning) need `drag` — `swipe`/`scroll` are direction-only and can't
   express them. Each end is independently a ref or `x,y`:
@@ -156,20 +166,29 @@ it rather than letting it appear unexplained in their `git status`.
   (`dpad_up`/`dpad_down`/`dpad_left`/`dpad_right`/`dpad_center`), or move
   focus directly with `tap-focused` (moves focus AND activates) / `focus`
   (moves focus only, no activation).
-- **Foldable (iPhone Duo)**: start on it by model name —
-  `vibeview dev --detach --json --model "iPhone Duo"` (MCP: `dev_start` with
-  `model`). Exact names come from `vibeview list-devices --models` (MCP:
-  `list_device_models`); a foldable is marked `foldable`. It starts closed, on
-  the cover screen. Fold with a preset or an exact hinge angle, and rotate:
+- **Foldables**: any model `vibeview list-devices --models` (MCP:
+  `list_device_models`) marks form factor `foldable` — the iPhone Duo,
+  Android foldables such as the Pixel 9 Pro Fold. Start on one by its exact
+  model name from that list, e.g.
+  `vibeview dev --detach --json --model "iPhone Duo"` or
+  `vibeview dev --platform android --detach --json --model "Pixel 9 Pro Fold"`
+  (MCP: `dev_start` with `model`). It starts closed, on the cover screen. Fold with a preset
+  or an exact hinge angle, and rotate:
   ```bash
   vibeview set-posture open --session <id>        # closed | partial | open
   vibeview set-posture --angle 75 --session <id>  # 0 (shut) to 180 (flat)
-  vibeview rotate --session <id>                  # quarter turn clockwise
-  vibeview rotate --degrees 270 --session <id>    # quarter turn back (180: half)
+  vibeview rotate --session <id>                  # Duo: quarter turn clockwise
+  vibeview rotate --degrees 270 --session <id>    # Duo: quarter turn back (180: half)
   ```
   Each response says where the device ended up — posture, angle and which
-  screen is lit (`cover` or `inner`; the device decides — a small angle can
-  keep the inner screen lit on the way down), or the new orientation. Folding
+  screen is lit (`cover` or `inner`; the device decides — on the Duo a small
+  angle can keep the inner screen lit on the way down), or the new
+  orientation. The posture names come from each device's own angle ranges,
+  so trust the reported `posture` over the angle you asked for: the Duo
+  reads 120-169 as `partial`, 170 up as `open`, and 1-119 by direction of
+  travel; an Android foldable uses its own ranges (the Pixel 9 Pro Fold: below
+  30 `closed`, 30-149 `partial`, 150 up `open`) and its inner screen whenever
+  it is not closed. An Android foldable rotates like a phone. Folding
   switches screens of different sizes and rotating turns the picture, so run
   `ui-tree` afterwards — old refs are stale. On other phones and tablets
   `rotate` toggles portrait/landscape (only 90 is accepted); TV devices don't
@@ -231,20 +250,20 @@ human-readable text.
 |------|------|---------|
 | `ui-tree` | — | Fetch the ref-annotated UI tree for the current screen. |
 | `logs` | `[--tail <n>] [--since <cursor>]` | Read recent app logs from the device (JS console, native errors, crash messages). The response's `cursor` feeds the next call's `--since`. |
-| `screenshot` | `[--out <path>]` | Capture a screenshot of the current screen and save it to disk. |
-| `tap` | `<target>` | Tap an element by ref (e.g. `@e5`), or raw coordinates (e.g. `100,200`). |
+| `screenshot` | `[--out <path>]` | Capture a screenshot of the current screen and save it to disk. On a phone or tablet its pixels are tap coordinates: a point read off it can be tapped as is — never rescale. |
+| `tap` | `<target>` | Tap an element by ref (e.g. `@e5`), or coordinates (e.g. `100,200`) in the space of the tree's rects and the screenshot's pixels. |
 | `long-press` | `<ref>` `[--ms <n>]` | Long-press (touch and hold) an element by its ref. |
 | `swipe` | `<direction>` `[--distance <short\|medium\|long>]` | Perform a scrollbar-semantic swipe gesture in one of four directions. |
 | `scroll` | `<direction>` | Scroll the current view up, down, left, or right using a gesture preset. |
 | `scroll-to` | `<text>` `[--direction <up\|down\|left\|right>] [--max-scrolls <n>] [--element-type <type>]` | Scroll until an element matching the text is visible. Stops at end of list. |
 | `drag` | `<from>` `<to>` `[--velocity <n>] [--press-ms <n>] [--hold-ms <n>]` | Drag between two points, each an `@ref` or `x,y`. The only precise two-point gesture. |
 | `alert` | `<get\|accept\|dismiss>` `[--button <label>]` | Inspect or respond to a system alert on iOS or Apple TV. |
-| `type` | `<text>` | Type text into the currently focused input field. |
+| `type` | `<text>` | Type text into the currently focused input field. On a phone or tablet a line break presses Enter and submits the field (`$'query\n'`), like `press enter`. |
 | `clear-text` | — | Clear the text in the currently focused input field. |
 | `press` | `<button>` | Press a device button or perform a system gesture (home, back, d-pad, etc). |
-| `open-url` | `<url>` | Open a deep link or URL in the app under test. |
-| `set-posture` | `<closed\|partial\|open>` or `--angle <0-180>` | Fold or unfold a foldable device (iPhone Duo) to a preset or an exact hinge angle — exactly one. Reports posture, angle and lit screen. Errors on a device without a hinge. |
-| `rotate` | `[--degrees <90\|180\|270>]` | Rotate the device and report the new orientation. Phones/tablets toggle portrait/landscape (90 only); a foldable turns a quarter clockwise by default. Not on TV. |
+| `open-url` | `<url>` | Open a deep link or URL in the app under test. The link goes to that app only; a link it has no screen for fails (no browser). |
+| `set-posture` | `<closed\|partial\|open>` or `--angle <0-180>` | Fold or unfold a foldable device (form factor `foldable`: the iPhone Duo, Android foldables) to a preset or an exact hinge angle — exactly one. Reports posture, angle and lit screen. Errors on a device without a hinge. |
+| `rotate` | `[--degrees <90\|180\|270>]` | Rotate the device and report the new orientation. Phones/tablets and Android foldables toggle portrait/landscape (90 only); the iPhone Duo turns a quarter clockwise by default. Not on TV. |
 | `wait` | `[--ms <n>]` | Wait for a specified number of seconds before continuing (default: 2s). |
 | `find` | `<text>` `[--below <text>] [--above <text>] [--near <text>]` | Find an element by text with optional spatial constraints. |
 | `tap-focused` | `<ref>` | TV: move focus to an element and press SELECT in one step. |