chore: release v0.10.0 (#164)

* docs(mcp): document --launch-timeout flag in run_tests tool description

* chore: release v0.10.0

Consolidates this release's CHANGELOG.md entries under a single
[0.10.0] section (was scattered across per-PR [Unreleased] appends),
bumps probe_agent to 0.10.0 (pubspec.yaml + agent_version.dart, kept
in sync per the comment in that file), vscode/package.json to 0.10.0,
and docs/wiki/Home.md's stated current version.

Covers PT-01 through PT-19 from IMPROVEMENT_TASKS.md: 15 real bugs
found and fixed (several with a different or larger root cause than
originally reported, confirmed via real-device verification), one new
config option (agent.launch_timeout / --launch-timeout), and a design
proposal for native-UI bridging (not implemented this release).

Author: pbertsch

CHANGELOG.md

@@ -18,7 +18,7 @@ Welcome to the FlutterProbe wiki. This documentation covers architecture details
 
 ## Project Status
 
-FlutterProbe is in active development. Current version: **0.9.9**.
+FlutterProbe is in active development. Current version: **0.10.0**.
 
 ### Repository Structure
 

docs/wiki/Home.md

@@ -204,6 +204,8 @@ Named parameters cover the most common options. Any probe test flag can be passe
 
 Key flags for the flags parameter:
   --timeout 60s               per-step timeout (default 30s)
+  --launch-timeout 3m         timeout for restart the app/clear app data specifically (default 120s) -
+                               raise this if those steps time out on an app with an expensive cold-launch path
   --format json|junit         structured output (default: terminal)
   --dry-run                   validate syntax without connecting to a device
   --parallel                  distribute tests across all connected devices

internal/mcp/server.go

@@ -1,15 +1,44 @@
 # Changelog
 
-## 0.9.10 - 2026-07-05
+## 0.10.0 - 2026-07-05
+
+A hardening release — every fix below was found and verified against a real
+device, working through a backlog of real E2E test issues surfaced by driver
+projects. Full detail on each is in the root CHANGELOG's `[Unreleased]`
+section (also being cut as part of this release); this is the summary
+relevant to apps embedding this package directly.
 
 - `probe.ping` now returns `agent_version` (this package's version) alongside
   `ok`, and accepts an optional `client_version` field from the CLI. Part of
-  the CLI↔agent version-compatibility handshake (see root CHANGELOG PT-07).
-  Both fields are additive and ignored by older CLIs/agents that don't know
-  about them.
+  the CLI↔agent version-compatibility handshake. Both fields are additive and
+  ignored by older CLIs/agents that don't know about them.
 - Fixed: the version reported in mDNS advertisements and `GET /probe/status`
   had drifted to `0.7.0` while this file's version moved on — corrected to
   match `pubspec.yaml` and moved into its own `agent_version.dart` file.
+- Fixed: `scroll`/`swipe` could report success while producing zero visible
+  movement (a single-jump synthetic drag, and stale content from a route
+  mounted underneath the current one both resolving as visible).
+- Fixed: `tap #id` could leave a text field genuinely unfocused, and the
+  `focused` state check (`see`/`don't see #id is focused`) had a false
+  positive matching ancestors of the selected element.
+- Fixed: `tap #id`'s fast direct-invoke path now recognizes `InkResponse`
+  (not just `InkWell`), covering more Material buttons before falling back
+  to a real hit-tested pointer tap.
+- Fixed: `wait until #id appears`/`disappears` always searched for the
+  literal text `"#my_button"` instead of resolving the id — timed out on
+  every non-text widget (icon buttons, etc.) regardless of visibility.
+- Fixed: `close keyboard` and `close the app` were both complete no-ops.
+  `close keyboard` now calls `FocusManager.instance.primaryFocus?.unfocus()`
+  directly; `close the app` calls `SystemNavigator.pop()`.
+- Fixed: `don't see #id is <state>` (negated state checks — `focused`,
+  `enabled`, `disabled`, `contains`) silently ignored the checked state,
+  degrading to a bare existence check.
+- Fixed: `scroll` could lose the gesture arena to `Dismissible`-wrapped list
+  rows and never actually scroll — now drives the nearest `Scrollable`'s own
+  `ScrollPosition` directly instead of simulating a pointer gesture.
+- Fixed: `take screenshot` could capture stale content from a previous route
+  instead of the current screen (missing `waitForSettled()` call, and
+  `_captureViaRepaintBoundary` had no route-awareness).
 
 ## 0.9.9 - 2026-05-13
 

probe_agent/CHANGELOG.md

@@ -9,4 +9,4 @@
 /// Lives in its own file (rather than server.dart, where it previously
 /// lived) so executor.dart can read it too without an executor.dart <->
 /// server.dart import cycle.
-const String probeAgentVersion = '0.9.9';
+const String probeAgentVersion = '0.10.0';

probe_agent/lib/src/agent_version.dart

@@ -3,7 +3,7 @@ description: >-
   On-device E2E test agent for FlutterProbe. Embeds in your Flutter app and
   executes test commands via direct widget-tree access with sub-50ms latency.
 
-version: 0.9.9
+version: 0.10.0
 homepage: https://flutterprobe.dev
 repository: https://github.com/AlphaWaveSystems/flutter-probe
 issue_tracker: https://github.com/AlphaWaveSystems/flutter-probe/issues

probe_agent/pubspec.yaml

@@ -2,7 +2,7 @@
   "name": "flutterprobe",
   "displayName": "FlutterProbe",
   "description": "High-performance E2E testing for Flutter apps — ProbeScript language support, local & cloud device testing (BrowserStack, Sauce Labs, AWS Device Farm, Firebase Test Lab, LambdaTest), visual regression, test recording, and Studio integration",
-  "version": "0.9.9",
+  "version": "0.10.0",
   "publisher": "flutterprobe",
   "icon": "resources/probe-icon.png",
   "engines": { "vscode": "^1.85.0" },

vscode/package.json

@@ -58,6 +58,7 @@ The `run_tests` tool has named parameters for common options (`paths`, `tag`, `d
 | Flag | What it does |
 |---|---|
 | `--timeout 60s` | Per-step timeout (default 30s) |
+| `--launch-timeout 3m` | Timeout for `restart the app`/`clear app data` specifically (default 120s) — raise this if those steps time out on an app with an expensive cold-launch path (e.g. Firebase App Check init) |
 | `-v` | Verbose step output — prints `→ step` before each step and `✓/✗ step (Xs)` after; slow steps (>5s) emit `⏱` progress ticks and a `⚠` warning at 80% of timeout. Recommended when diagnosing failures. |
 | `--format json` | Structured JSON results (pipe to `get_report`) |
 | `--format junit` | JUnit XML for CI systems |