Add GUI test plan

.gitea/workflows/ci.yml

@@ -8,6 +8,9 @@ on:
       - "v*"
       - "release-*"
       - "[0-9]+.[0-9]+.[0-9]+*"
+  pull_request:
+    branches:
+      - main
 
 permissions:
   contents: write
@@ -16,7 +19,6 @@ env:
   GO_VERSION: "1.26.1"
   ANDROID_SDK_ROOT: /opt/android-sdk
   ANDROID_NDK_ROOT: /opt/android-sdk/ndk
-  JAVA_HOME: /usr/lib/jvm/java-21-openjdk-amd64
   DIST_DIR: dist
 
 jobs:
@@ -31,6 +33,12 @@ jobs:
         with:
           go-version: ${{ env.GO_VERSION }}
 
+      - name: Setup Java
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: "25"
+
       - name: Install native build dependencies
         shell: bash
         run: |
@@ -78,6 +86,12 @@ jobs:
         with:
           go-version: ${{ env.GO_VERSION }}
 
+      - name: Setup Java
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: "25"
+
       - name: Install native build dependencies
         shell: bash
         run: |

AGENTS.md

@@ -177,7 +177,7 @@ These features are product requirements, not “nice to have” ideas.
   local `ANDROID_NDK_ROOT=/opt/android-ndk`,
   CI `ANDROID_NDK_ROOT=/opt/android-sdk/ndk`,
   local `JAVA_HOME=/usr/lib/jvm/java-25-openjdk`,
-  CI `JAVA_HOME=/usr/lib/jvm/java-21-openjdk-amd64`.
+  CI `JAVA_HOME` provided by `actions/setup-java` with Temurin 25.
 - Remember the known Android runtime regression:
   `gioui.org v0.9.0` produced a black screen on the `KeepassGoAPI35` emulator, while `gioui.org v0.8.0` rendered correctly. Treat Gio upgrades on Android as regression-sensitive and verify them on-device or in the emulator.
 - When validating an APK in the emulator, prefer the known KeePassGO setup:

APK.md

@@ -15,7 +15,8 @@ make apk-release
 `make apk` uses a local Java 25 install when `JAVA_HOME` points to one.
 If the host does not have a working Java 25 install, it falls back to the
 repo-managed Docker image in `packaging/docker/android-apk/`, which also builds
-with Java 25.
+with Java 25. CI provisions Java 25 directly in the build job so release builds
+use that same local path instead of nested Docker.
 
 `make apk` remains a developer build path and may use Gio's default debug or
 ephemeral signing behavior if no explicit signing key is provided.
@@ -67,7 +68,7 @@ The Android build uses the branded icon asset at:
 
 Note:
 
-- KeePassGO's documented Android build uses Java 25 locally.
+- KeePassGO's documented Android build uses Java 25 locally and in CI.
 - If that host setup is unavailable, `make apk` falls back to the Docker image
   so the build still runs under Java 25 instead of encoding a newer host JDK as
   a requirement.

Makefile

@@ -86,7 +86,7 @@ apk-local: android/keepassgo-android.jar
 apk-release:
 	@test -f "$(RELEASE_SIGNKEY)" || { echo "Release signing key not found at $(RELEASE_SIGNKEY)"; exit 1; }
 	@test -f "$(RELEASE_SIGNPASS_FILE)" || { echo "Release signing password file not found at $(RELEASE_SIGNPASS_FILE)"; exit 1; }
-	@$(MAKE) apk SIGNKEY="$(abspath $(RELEASE_SIGNKEY))" SIGNPASS_FILE="$(abspath $(RELEASE_SIGNPASS_FILE))"
+	@$(MAKE) apk SIGNKEY="$(abspath $(RELEASE_SIGNKEY))" SIGNPASS_FILE="$(abspath $(RELEASE_SIGNPASS_FILE))" JAVA_HOME="$(JAVA_HOME)"
 
 apk-container: apk-container-image
 	@command -v docker >/dev/null 2>&1 || { echo "docker is required for apk-container"; exit 1; }

README.md

@@ -95,8 +95,9 @@ make apk
 
 `make apk` prefers a local Java 25 install at `JAVA_HOME`. If that is not
 available, it falls back to the repo-managed Docker build image, which also
-uses Java 25. You still need the Android SDK and NDK installed and configured
+uses Java 25. CI provisions Java 25 directly in the build job so release
-for real device or release packaging.
+packaging follows that same local path. You still need the Android SDK and NDK
+installed and configured for real device or release packaging.
 
 Release package:
 

docs/gui-test-plan.md

@@ -0,0 +1,207 @@
+# GUI Test Plan
+
+This document splits GUI validation into human-owned and agent-owned coverage.
+The intent is that, together, both passes exercise every reachable GUI area
+without relying on one side to judge both functional correctness and real
+usability.
+
+## Scope
+
+- Desktop GUI
+- Android GUI
+- Shared workflow surfaces:
+  lifecycle, unlock, list/detail/editor, settings, sync, API tokens, API audit,
+  templates, recycle bin, About, and platform-specific integrations
+
+## Ownership Model
+
+- Human testing covers visual correctness, ergonomics, discoverability, focus,
+  scrolling, tap targets, keyboard feel, and platform integration behavior.
+- Agent testing covers deterministic automation:
+  build, install, launch, focus, test suite, lint, and release-path validation.
+
+## Android Plan
+
+### Coverage Matrix
+
+| Area | Human | Agent |
+|---|---|---|
+| App launch and first render | Verify real rendering, layout, readability, no black screen | Verify emulator online, app installed, app focused |
+| Lifecycle screen | Exercise create/open/open-remote forms visually | Cover lifecycle logic through Go tests and release APK build |
+| Unlock flow | Verify password/key-file affordances, focus, error messaging | Covered by automated tests |
+| Vault list and navigation | Verify scrolling, selection, section switching, phone layout reachability | Covered by automated tests for list state/search/section behavior |
+| Entry detail | Verify readability, action placement, tap targets | Covered by automated tests for state transitions and copy/generate actions |
+| Entry editor | Verify field editing usability and save affordance | Covered by automated tests for save/update behavior |
+| Template views | Verify template list/editor reachability and copy | Covered by automated tests |
+| Recycle bin | Verify deleted-entry browsing/search UX | Covered by automated tests |
+| Attachments UI | Verify list/add/remove affordances visually | Covered by automated tests for attachment summaries and save paths |
+| Search | Verify placeholder, live filtering feel, clear behavior | Covered by automated tests for search semantics |
+| Settings | Verify toggles and summaries are understandable | Covered by automated tests for persistence |
+| Remote sync setup/dialog | Verify sheet/dialog layout, field order, source/direction controls | Covered by automated tests and release build |
+| Saved remote binding UI | Verify summaries are understandable and discoverable | Covered by automated tests |
+| API tokens / API audit | Verify navigation and dense-detail readability | Covered by automated tests |
+| About / informational screens | Verify copy, scroll, layout | Covered by automated tests for search disable state and section state |
+| Android-only share/import/file picker | Verify system integration actually works | Agent only verifies package/build/focus, not picker UX |
+| Android autofill entry point | Verify Android mechanism appears and is usable | Agent only verifies app/package/focus |
+
+### Human Steps
+
+1. Start from the currently installed `org.julianfamily.keepassgo` app on the
+   running emulator or device.
+2. Verify launch lands on a rendered screen, not a black frame, blank frame, or
+   frozen frame.
+3. On the lifecycle screen, inspect every visible action:
+   `Create vault`, `Open vault`, `Open remote vault`, unlock-related controls,
+   recent vaults, and any sync summaries.
+4. Create a throwaway vault or open a demo vault and verify the full unlock path
+   is visually clear.
+5. After unlock, visit each top-level section you can reach in the Android UI:
+   vault/group list, entry detail, entry editor, templates, recycle bin,
+   settings, API tokens, API audit, About.
+6. In the main vault UI, verify:
+   group navigation, entry selection, search, back behavior, scrolling, and any
+   phone-only toggles or drawers.
+7. Open an entry and exercise every visible entry action:
+   copy username, copy password, copy URL, reveal/hide password, generate
+   password, save.
+8. Exercise one template flow:
+   open template list, inspect template detail/editor, save or cancel.
+9. Exercise recycle-bin browsing and search.
+10. Open settings and inspect every visible toggle or summary card. Change at
+    least one benign toggle and verify it sticks after leaving and returning.
+11. Open the sync UI and inspect:
+    remote setup, saved-binding summary, direction/source choices,
+    confirm/cancel behavior, and text summaries.
+12. Exercise one Android-native integration:
+    shared-vault import, file picker open, or current-vault share.
+13. If Android autofill is in scope for current testing, open a target app/site
+    and confirm the Android autofill entry point is offered and usable.
+14. Record failures with:
+    exact screen, exact control, expected behavior, actual behavior, and whether
+    it is Android-only or likely shared with desktop.
+
+### Agent Steps Executed
+
+1. Verified emulator availability with `adb devices -l`.
+2. Verified emulator Android version with
+   `adb shell getprop ro.build.version.release`.
+3. Verified app package installed with
+   `adb shell pm list packages org.julianfamily.keepassgo`.
+4. Verified app focus with
+   `adb shell dumpsys window | rg 'mCurrentFocus|mFocusedApp'`.
+5. Ran `go test ./...`.
+6. Ran `go tool golangci-lint run ./...`.
+7. Verified release APK path through release workflow `v0.6.0`:
+   PR run `60` succeeded,
+   post-merge `main` run `61` succeeded,
+   tag run `62` succeeded.
+
+### Agent Results
+
+- Emulator present: yes
+- Android version: `15`
+- App installed: yes
+- App focused:
+  `org.julianfamily.keepassgo/org.gioui.GioActivity`
+- APK release build path: passed in release CI
+- Remaining Android risk:
+  real visual usability and system-integration behavior
+
+## Desktop Plan
+
+### Coverage Matrix
+
+| Area | Human | Agent |
+|---|---|---|
+| App launch and window presentation | Verify startup polish, focus, resize behavior, DPI/readability | Verify desktop binaries built in release CI |
+| Lifecycle screen | Verify desktop layout density, affordance clarity, recent vault usability | Covered by automated tests |
+| Unlock flow | Verify keyboard-first behavior, focus order, errors, lock/unlock loop | Covered by automated tests plus human keyboard feel |
+| Group browser and list pane | Verify density, selection clarity, scrolling, split behavior | Covered by automated tests for state/search |
+| Entry detail pane | Verify copy actions, reveal/hide, path context, readability | Covered by automated tests |
+| Entry editor | Verify field order, keyboard flow, save/cancel usability | Covered by automated tests |
+| Templates | Verify templates section is reachable and understandable | Covered by automated tests |
+| Recycle bin | Verify browsing, searching, deletion or recovery UX | Covered by automated tests |
+| Search | Verify keyboard search flow, placeholder correctness, result context | Covered by automated tests |
+| Vault save/save-as/lock | Verify menus, buttons, and shortcuts feel correct | Covered by automated tests and release builds |
+| Settings | Verify desktop layout, summaries, toggles, persistence feel | Covered by automated tests |
+| Remote sync UI | Verify dialog layout, wording, discoverability, advanced options | Covered by automated tests |
+| API tokens | Verify dense-detail presentation and policy editor usability | Covered by automated tests |
+| API audit | Verify search/filter/readability | Covered by automated tests |
+| Browser-extension-adjacent desktop UX | Verify visible status/help text and extension workflow discoverability | Agent validated release/build path; human should judge usability |
+| About and docs entry points | Verify copy and layout | Covered partly by tests; human judges presentation |
+
+### Human Steps
+
+1. Launch KeePassGO on desktop from the shipped binary or normal desktop entry.
+2. Inspect the lifecycle/open screen at normal window size and at a narrower
+   width.
+3. Exercise create, open, save-as, lock, unlock, and reopen on a throwaway
+   vault or demo vault.
+4. Use keyboard-first operation for at least one complete pass:
+   tab order, enter or escape expectations, search focus, unlock focus, and
+   editor save flow.
+5. After unlocking, visit every reachable primary section:
+   vault list, entry detail, editor, templates, recycle bin, settings,
+   API tokens, API audit, About.
+6. In the vault browser, verify:
+   nested groups, path context, scrolling, selection state, and search results
+   with path context.
+7. Open an entry and exercise all visible actions:
+   copy username, password, URL, reveal/hide password, password generation,
+   and save.
+8. Exercise at least one mutation flow:
+   create entry, edit entry, move/delete entry, view recycle bin, and recover
+   if available.
+9. Open settings and inspect all visible summaries and toggles.
+10. Open remote sync UI and inspect every visible mode:
+    setup, saved binding, advanced sync, source or direction choices.
+11. Open API Tokens and API Audit even if you do not issue a real token, just
+    to assess navigation and readability.
+12. If you use browser integration, verify the desktop-side flow is still
+    understandable from the product UI and extension behavior.
+13. Record failures with:
+    screen, control, expected behavior, actual behavior, and whether it is
+    presentation-only or functional.
+
+### Agent Steps Executed
+
+1. Verified clean repo state on `main`.
+2. Ran `go test ./...`.
+3. Ran `go tool golangci-lint run ./...`.
+4. Verified post-merge `main` CI run `61` succeeded.
+5. Verified release tag run `62` succeeded.
+6. Verified release `v0.6.0` published.
+7. Verified release artifacts include:
+   `keepassgo-linux-amd64`,
+   `keepassgo-windows-amd64.exe`,
+   `keepassgo-windows-arm64.exe`,
+   and `keepassgo.apk`.
+
+### Agent Results
+
+- Automated logic/state coverage: pass
+- Desktop build coverage: pass
+- Release publication: pass
+- Remaining desktop risk:
+  interaction quality, keyboard feel, dense-layout readability, and workflow
+  discovery
+
+## Reporting Template
+
+Use this format when reporting findings from the human pass:
+
+```md
+## Android
+- Screen:
+- Action:
+- Expected:
+- Actual:
+- Severity:
+
+## Desktop
+- Screen:
+- Action:
+- Expected:
+- Actual:
+- Severity:
+```