// Audit translation accuracy for strings changed in a GitHub PR. Use when asked to check, review, audit, or verify translations in a pull request. Identifies inaccurate translated strings by comparing PR changes against the English source.
Add UI automation tests to Trail-Sense Android app using AutomationLibrary. Use when asked to create, add, write, or implement automated tests, UI tests, integration tests, or androidTests for Trail Sense tools. Covers test class structure, AutomationLibrary functions, and testing patterns.
Add groupable support to a Trail Sense feature or model. Use when asked to make an item groupable, add folders/groups, add group navigation, or retrofit an existing list using IGroupable/Groupable, GroupLoader, GroupListManager, group pickers, and Room parent/group persistence.
Add new Room database persistence to Trail-Sense Android app. Use when the user asks to create, add, or implement database persistence for a model, including Entity, DAO, Repository, and AppDatabase migration. Covers entity-to-model mapping, index configuration, and standard CRUD operations.
Write or update user guide documentation for Trail-Sense tools. Use when asked to create, write, update, or modify a tool's user guide, help documentation, or user-facing documentation. Covers guide structure, formatting conventions, and common sections.
| name | android-check-pr-translations |
| description | Audit translation accuracy for strings changed in a GitHub PR. Use when asked to check, review, audit, or verify translations in a pull request. Identifies inaccurate translated strings by comparing PR changes against the English source. |
Audit the translation accuracy of Android string resources changed in a GitHub pull request.
Run the extraction script from the skill's folder to get the changed strings from the PR:
python scripts/extract_pr_strings.py <pr-number>
The output includes both the translated value and the English source for each changed key — no repo search needed.
For each entry in added, compare value (translation) against english (source) using the accuracy criteria below.
Report findings as JSON (see Output Format).
Counts as inaccurate:
%s, %1$d, %2$s) are missing, reordered incorrectly, or alteredDoes NOT count as inaccurate:
scripts/extract_pr_strings.py (skill-local)The script parses the output of gh pr diff and extracts <string> elements from res/values-*/strings.xml patch hunks.
Usage:
python scripts/extract_pr_strings.py <pr-number>
Output schema:
{
"pr_number": 1234,
"locales": {
"values-pl": {
"added": [{"key": "some_key", "value": "translated text", "english": "English source text"}],
"removed": [{"key": "old_key", "value": "old translated text", "english": "English source text"}]
}
}
}
added — strings newly introduced or whose value was changed (new value)removed — strings that were deleted or whose value was changed (old value)english — the corresponding value from app/src/main/res/values/strings.xml; empty string if the key is not foundres/values-*/strings.xml files are included; the English res/values/strings.xml is excludedOutput ONLY a JSON blob:
{
"pr_number": 1234,
"inaccurate_translations": [
{
"key": "key_name",
"locale": "values-xx",
"reason": "Why it is inaccurate"
}
]
}
If all translations are accurate:
{
"pr_number": 1234,
"inaccurate_translations": []
}
Do not output any explanation, commentary, or text outside the JSON blob.