Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Quality Assurance Cycle

    1

    Shipping a feature doesn't mean it's actually finished.

    Free

    Security scanned
    Quality Assurance Cycle

    Quality Assurance Cycle

    Example session with this skill installed

    manual QA branch --platform android

    • Read your context and instructions
    • Compiled the quality assurance cycle

    I have audited the changes on the current branch against the Android Material 3 and Navigation bars. I ran the build and exercised the profile flow on a Pixel 9a emulator.

    QA — branch · 1 blocker · 2 major · 1 minor

    🔴 Blocker

    • EditProfileScreen.kt:88 — subtitle clips silently (no maxLines, parent height fixed at 120.dp). Add maxLines = 2, overflow = Ellipsis.

    🟠 Major

    • SettingsRow.kt:31 — Color(0xFF222222) divider invisible in dark mode. Use colorScheme.outlineVariant.
    • ProfileGraph.kt:57 — popUpTo(Login) should be popUpTo(Profile); back press bounces user to Login.

    🟡 Minor

    • ProfileCard.kt:12 — missing contentDescription on avatar image. Add stringResource(R.string.desc_avatar).

    ❓ Needs device verification

    • Theme.kt:45 — custom ripple alpha might be too faint on physical OLED screens under direct light.

    Tests: ./gradlew assembleDebug test lint — builds, 142 passed, lint clean.
    Manual: Pixel 9a emulator, edit-profile flow at font_scale 1.3 — screencap confirms the subtitle clip above.

    Next steps

    • Run /qa-cycle fix to automatically generate worktrees and apply these one-line fixes.
    • Use manual-qa last to check for regressions in the specific theme token changes.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Shipping a feature doesn't mean it's actually finished. Code reviews miss navigation traps, and unit tests can't see clipped UI or state survival issues after process death. This skill package provides a professional-grade QA department that maps your app, identifies regressions, and fixes bugs in isolated workstreams.

    What it does

    • Feature mapping builds a comprehensive map of every route, component, and shared module in your application to ensure full coverage.
    • Platform-specific auditing judges changes against the specific human interface guidelines for iOS, Android, and Web (WCAG 2.2 AA).
    • Blast-radius analysis greps the codebase for shared symbols to catch regressions in distant screens caused by theme or DTO changes.
    • Automated fix cycles fans out agent fixers to individual git worktrees to resolve triaged bugs without human intervention.
    • Verification loops assigns fresh "verifier" agents to prove every fix on a clean build of the main branch before closing an item.

    How it works

    1. Orchestrate a cycle using qa-cycle to map the app, plan small test units, and dispatch parallel testers to sandboxed environments.
    2. Review the triage list where findings are grouped by severity (Blocker, Major, Minor) with one-line symptom-and-fix descriptions.
    3. Execute fixes by authorizing the agent to fan out fixers and fresh verifiers until the "Definition of Done" criteria are met.
    4. Run targeted regressions via /regression or /manual-qa to verify a specific branch or diff against platform bars and blast-radius checks.

    Frameworks & tools

    Supports Android (Compose, Gradle), iOS (Swift, Xcode), and Web (React, Vue, Svelte, Tailwind). Integrates with Git worktrees, sandboxed emulators, and browsers for visual verification.

    Why this beats prompting it yourself

    Generic prompts fail to catch lifecycle issues like state restoration or cross-platform differences in back-stack behavior. This skill uses a multi-agent orchestration pattern that separates the fixer from the verifier, preventing the "confirmation bias" that occurs when a single agent checks its own work.

    Use cases

    • Auditing a complex UI refactor for accessibility and navigation regressions across Android and iOS.
    • Running a full QA cycle on a large pull request to catch edge cases like RTL layout breaks or font scaling issues.
    • Mapping an unfamiliar codebase to identify dead code and breaking changes in shared repository layers.
    • Closing a backlog of triaged "Minor" polish bugs and verifying them in parallel worktrees.

    Known limitations

    Requires a functional build environment for local testing. Visual verification depends on the availability of emulators, simulators, or browser drivers on the host machine.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Mobile Development.

    Requires functional local build environment. Visual verification depends on host-side emulators/browsers. Read-only by default until user confirms triage.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    • Free with an Agensi account

    Software
    Node.jseslinttscPlaywrightGit

    Permissions

    Terminal / Shell
    Browser
    Read Files
    Write Files
    Listed22 days ago
    Updated2 days ago

    What's inside

    Frequently Asked Questions

    The skill uses a multi-agent orchestration pattern where a 'fixer' agent resolves the bug in an isolated git worktree, and a completely separate 'verifier' agent attempts to break it on a fresh build. This eliminates confirmation bias, as the agent checking the code isn't the one that wrote it.

    qa-cycle is designed for cross-platform projects, supporting Android (Compose), iOS (Swift), and Web frameworks like React and Vue. It specifically audits against platform-specific standards, such as Human Interface Guidelines for mobile and WCAG 2.2 for web.

    You will receive the full skill package including the feature-mapping engine, blast-radius analyzer, and the multi-agent orchestration logic. It also includes pre-configured workflows for git worktree management and platform-specific auditing templates.

    The skill requires a functional build environment on the host machine, including the necessary compilers (Xcode/Gradle) and local emulators, simulators, or browser drivers for visual verification.

    Unlike generic prompts, this skill performs 'blast-radius analysis' to see how changes to shared modules or DTOs impact distant parts of the app, and it tests for lifecycle issues like state survival and back-stack behavior that basic LLM queries often miss.

    New in Mobile Development

    Free