菜单
Run the full FeedFlow Maestro release validation suite and produce HTML and Markdown pass/fail reports. Use when the user asks to run all Maestro tests, validate E2E coverage before release, check Android and iOS Maestro flows, or summarize failed FeedFlow E2E flows.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Update and localize FeedFlow release notes when the user provides platform-specific changelog copy. Use for Android Google Play, iOS App Store, macOS App Store, and Windows Microsoft Store release-note updates, including requests to translate notes for every configured store locale.
Audit FeedFlow localization, store listing, live App Store/Play Store listing state, and screenshot-copy changes. Use when checking whether new app translations were added or completed, whether store copy/localized listings changed, whether live store metadata is stale, or whether screenshots need to be regenerated or uploaded.
Generate a new localized set of FeedFlow App/Play Store screenshots in the Figma screenshot blueprint file by duplicating the existing ru-RU export (not the English baseline) and swapping text. Use when the user asks to localize store screenshots, generate screenshots for a new language, or export screenshots for a language other than English/Russian.
Investigate and fix FeedFlow crashes from Sentry triage notes. Use when the user provides a local note path, note URL, Gmail-exported note, Sentry issue URL, Sentry issue ID, or event ID and asks Codex to check why a crash is happening, explain the root cause, propose or implement a fix, validate it, commit it, or close/resolve the Sentry issue.
Update Google Play listing files (title, short description, full description) from storecopy source assets. Use when the user wants to update Play Store listings for a specific language or all languages, sync store copy to listing files, or refresh Play Store translations.
Generate release notes for app stores (Android Play Store, iOS App Store, macOS App Store, Linux Flatpak). Use when the user asks for changelogs, release notes, or store descriptions based on git history.
| name | run-maestro-release-tests |
| description | Run the full FeedFlow Maestro release validation suite and produce HTML and Markdown pass/fail reports. Use when the user asks to run all Maestro tests, validate E2E coverage before release, check Android and iOS Maestro flows, or summarize failed FeedFlow E2E flows. |
Run FeedFlow's full local Maestro release gate across Android and iOS, then report what passed, what failed, and where the logs are.
Confirm the repo root is the FeedFlow project containing e2e/maestro/ and e2e/scripts/.
Run the bundled script from the repo root:
python3 .ai/skills/run-maestro-release-tests/scripts/run_maestro_release_tests.py
If the user asks for a narrower pass, pass explicit flags:
# Android only
python3 .ai/skills/run-maestro-release-tests/scripts/run_maestro_release_tests.py --platform android
# iOS only
python3 .ai/skills/run-maestro-release-tests/scripts/run_maestro_release_tests.py --platform ios
# Smoke only while debugging
python3 .ai/skills/run-maestro-release-tests/scripts/run_maestro_release_tests.py --suite smoke
Open or read the generated report.html first. Use report.md as a terminal-friendly fallback. Summarize:
maestro.Resizable_Experimental.iPhone 17 Pro simulator. The script uses SIMULATOR_UDID when present, otherwise it detects the booted simulator.iosApp/FeedFlow.xcodeproj; the script generates it when missing.When both platforms are selected, the script runs Android and iOS in parallel. Each platform still builds and installs once, pushes fixtures, then runs every YAML flow for that platform in sorted order from:
e2e/maestro/android/smokee2e/maestro/android/regressione2e/maestro/ios/smokee2e/maestro/ios/regressionIt continues after failed flows so the final report captures the full failure set. It exits non-zero when any setup step or flow fails.
Reports and logs are written under:
.tmp/maestro-release-tests/<timestamp>/
The main visual report is report.html. It contains run metadata, setup/build steps, suite summaries, failed flows, brief failure classifications, evidence lines from logs, and every selected test with links to per-flow logs. The script also writes report.md for quick terminal review.
Use the per-flow logs in that directory together with Maestro artifacts under ~/.maestro/tests/ when debugging failures. Treat the report classification as a first-pass diagnosis, then verify ambiguous cases against screenshots and UI hierarchies before calling a failure a product regression.