agentsclimarketplace

Ios ai ui check

Skill Kofiloski/ios-release-quality-toolkit/plugins/ios-release-quality-toolkit/skills/ios-ai-ui-check

Agent skills for App Store review risk, XCUITest testability repairs, and evidence-rich iOS Simulator checks.

Install
npx -y skills add Kofiloski/ios-release-quality-toolkit --skill ios-ai-ui-check

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Add, operate, or debug reproducible iOS Simulator UI checks in GitHub Actions with ios-ai-ui-check. Use when someone asks for AI-planned or deterministic XCUITest checks from a plain-English goal, wants video/screenshots/logs/xcresult artifacts and a PR summary, needs to scaffold an app-owned runner or scenario, or is troubleshooting planner validation, Simulator, runner, artifact-manifest, or pull-request-comment failures. Keep app launch state, fixtures, accessibility identifiers, UI interactions, and assertions in the app repository; keep the reusable action provider-agnostic.

SKILL.md

3.5 KB, 602 tokens by cl100k_base, as published. Nobody here has run it

iOS AI UI Check

Integrate the public Kofiloski/ios-ai-ui-check action without hiding app-specific test truth inside the reusable workflow.

Workflow

  1. Read the target repository's instructions and CI setup. Identify the .xcodeproj, shared scheme, UI-test target, supported Simulator destination, and current launch/testing helpers.
  2. Choose the least-variable mode that satisfies the request:
    • Prefer a checked-in scenario for required or security-sensitive CI.
    • Use a planner goal when exploratory coverage is useful and the app repo can explicitly control the context sent to its chosen provider.
  3. Read references/adoption.md before scaffolding or editing a workflow.
  4. Pin the public action to a release tag or commit SHA. Do not track main in stable CI.
  5. Use the upstream scaffold for a new integration. Run its dry-run first, review the proposed files, then generate them only within the target app repo.
  6. Review the generated app-owned contract before running it:
    • deterministic launch arguments and environment values
    • safe fixtures, mocks, or test accounts
    • stable accessibility identifiers on the actual controls
    • focused assertions that prove the requested behavior
    • planner context that contains no secrets
  7. Run the checked-in scenario locally before enabling a PR workflow. Treat Simulator availability, Xcode version, and shared-scheme failures as setup problems rather than planner problems.
  8. Inspect summary.md, manifest.json, logs, screenshots, video, and .xcresult together. Do not report success from a status dot when the evidence contradicts it.
  9. Keep fixes at the correct boundary:
    • reusable orchestration or artifact-contract defect -> fix ios-ai-ui-check
    • element lookup, wrong element type, identifier collision, or unreachable route -> use $ios-ui-testability-contract
    • app launch, data, interaction, or assertion logic -> fix the app-owned runner

Guardrails

  • Do not move app-specific launch, seeding, UI interaction, or assertions into the reusable action.
  • Do not add a hard-coded model provider to the action. Only the app-owned planner command may choose a provider.
  • Never pass secrets through prompts, planner context, logs, scenarios, screenshots, or artifacts.
  • Prefer pull_request when compiling and running pull-request code. Do not switch to pull_request_target to gain secret access.
  • Start with manual or label-triggered runs. Make the check required only after launch state and identifiers are deterministic.
  • Do not promise visual correctness unless the app-owned scenario contains an explicit, reviewable assertion for it.

Completion evidence

Report:

  • selected deterministic or planner mode and why
  • action ref and Simulator/Xcode assumptions
  • app-owned files added or changed
  • exact local or CI command run
  • artifact paths inspected
  • any remaining app-specific contract or credential work

What ships with it: 2 files

2.9 KB alongside SKILL.md

agents/

references/

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.