SwiftUI Pro
Review SwiftUI code for modern APIs, maintainability, accessibility, and performance problems.
The skill is limited to SwiftUI code review, declares no network, file-write, or other external side effects, and says not to add third-party frameworks without asking first, reducing permission and dependency risk. Points are deducted because it does not define sensitive-data handling, data-flow disclosure, user-confirmation boundaries, rollback, or recovery after failure; author metadata and the MIT license provide partial attribution, while publisher identity remains unverified.
The review steps, core rules, and output structure are broadly consistent, and the referenced materials cover APIs, views, data, navigation, design, accessibility, performance, Swift, and hygiene. Points are deducted because those reference files are not included in the supplied material, so dependency availability, rule consistency, key-path reproduction, abnormal-input handling, and failure feedback cannot be verified; static calibration also limits the score.
The intended audience and scenarios are reasonably clear: reading, writing, and reviewing SwiftUI projects, with support for focus-specific partial reviews. Points are deducted because non-fit boundaries, precise trigger conditions, input prerequisites, and Chinese-language support are not specified; the iOS 26 and Swift 6.2 defaults may not fit legacy projects or non-Apple environments.
SKILL.md includes a name, description, argument hint, version, author, license, layered review process, partial-review guidance, and output example; repository context adds installation, contribution, and MIT-license information. Points are deducted because reference-file installation or existence is not self-verified in the selected file, and known limitations, troubleshooting, changelog, explicit maintenance responsibility, and update path are incomplete.
The required output—file-based findings with rules, before/after fixes, and a prioritized summary—is directly usable, and the stated task matches the claimed value. Points are deducted because no real review outputs, test suite, or third-party execution evidence are supplied, so correctness, coverage, and marginal benefit over manual review cannot be confirmed; static calibration caps effectiveness at 7.
The skill makes auditable claims about review categories and output format, while the README supplies installation and invocation examples. Points are deducted because the reference contents, tests, CI validation, and independent reproduction evidence are absent; key claims therefore remain primarily author assertions, warranting only limited static verifiability.
- The referenced ${CLAUDE_SKILL_DIR}/references/*.md files were not supplied; after installation, verify that they exist and match SKILL.md.
- The iOS 26 and Swift 6.2 defaults may produce unsuitable advice for legacy projects; confirm the project's deployment targets first.
- The skill does not define handling boundaries for sensitive code, private data, network dependencies, or external tools, and provides no failure-recovery or rollback guidance.
What does this skill do, and when should you use it?
SwiftUI Pro is an Agent Skill for reviewing Swift and SwiftUI projects in Claude Code, Codex, and other AI coding tools. It checks deprecated APIs, views and animations, data flow, navigation, design, accessibility, performance, modern Swift, and code hygiene. Its guidance targets iOS 26 and later and Swift 6.2 or later. It fits developers who want structured, Apple-oriented review feedback, although the source provides no test suite or project-specific validation results.
It reads Swift and SwiftUI code and, when relevant, loads reference files covering APIs, views, data flow, navigation, design, accessibility, performance, Swift, and hygiene. It does not describe running builds, tests, or network calls. Instead, it reports genuine issues organized by file, identifies relevant lines and violated rules, shows brief before-and-after fixes, and ends with a prioritized summary.
- An iOS developer building an iOS 26+ SwiftUI app who wants to check modern API usage.
- A team reviewing AI-generated SwiftUI changes for deprecated APIs, view structure, and animation choices.
- An app developer validating Dynamic Type, VoiceOver, Reduce Motion, and accessible labels for icon-only controls.
- A developer investigating state management, shared data flow, or navigation performance issues.
- A SwiftUI team that wants review guidance without introducing UIKit or third-party frameworks.
What are this skill's strengths and limitations?
- Covers API modernization, views, data flow, navigation, design, accessibility, performance, and modern Swift.
- Explicitly asks the reviewer to report genuine problems rather than inventing or nitpicking issues.
- Uses file-based findings with named rules and before-and-after examples.
- Targets iOS 26+ and Swift 6.2+ with modern Swift concurrency guidance.
- MIT licensed for commercial use, modification, distribution, and private use.
- The source provides no automated tests, build validation, or project-level execution results.
- Its scope is Swift and SwiftUI review; sibling SwiftData, Swift Concurrency, and Swift Testing skills should not be attributed to it.
- Reference paths use the CLAUDE_SKILL_DIR variable, so other clients must provide an equivalent skill-directory context.
- The README does not provide a complete client compatibility matrix or detailed failure and permission behavior.
How do you install this skill?
Install it with npx:
npx skills add https://github.com/twostraws/swiftui-agent-skill --skill swiftui-pro
If npx is unavailable, the README recommends installing Node with brew install node; Homebrew may need to be installed first. You can also clone the repository and install it manually. Claude Code users can additionally run /plugin marketplace add twostraws/SwiftUI-Agent-Skill followed by /plugin install swiftui-pro@swiftui-agent-skill.
How do you use this skill?
After installation, trigger SwiftUI Pro with /swiftui-pro in Claude Code or $swiftui-pro in Codex. You can provide a focus area, such as “$swiftui-pro Focus on accessibility,” or use natural language to request a performance review. The SKILL.md exposes a focus-area argument hint; exact installation locations and trigger syntax for every other client are not documented in the source.